Consolidate agents into unified pipelines (shuri, vision, friday, bifrost)#50
Open
Consolidate agents into unified pipelines (shuri, vision, friday, bifrost)#50
Conversation
- Shuri (01): spec-writer + todo-generator + sync-local + kanban-converter → Unified doc pipeline (spec → todo Kanban → sync locale) - Vision (05): code-auditor + code-simplifier → Code audit (8 dimensions) + simplification via /simplify - Friday (09): context-generator + documentalist → Context snapshot (15K chars) + /docs organization & health - Bifrost (21): notion-importer + beetlejuice → Notion bridge (import pages + export QA with User Stories) Update CLAUDE.md and agents/CLAUDE.md with new references. https://claude.ai/code/session_01Sz6RTvd8y7ccHjuVsAmDsp
Replace all references to the 10 deleted agents (spec-writer, todo-generator, sync-local, code-auditor, code-simplifier, context-generator, documentalist, notion-importer, beetlejuice, kanban-converter) with their 4 merged successors (shuri, vision, friday, bifrost) across: - install.sh: usage section with new agent names - site/data/commands.json: agent entries and workflows - site/index.html: agentFileMap - commands/agents/: remove broken symlinks, add new ones - commands/README.md: agent table and orchestrator workflows - README.md: agent table, native features, workflows - agents/Readme.md: directory listing and workflow diagrams - CLAUDE.md: external state reference - .claude/rules/native-features.md: agent references https://claude.ai/code/session_01Sz6RTvd8y7ccHjuVsAmDsp
📝 WalkthroughWalkthroughThis PR consolidates 10 single-purpose agents into 4 multi-mode agents—shuri (documentation pipeline), vision (code audit & simplification), friday (docs management & context), and bifrost (Notion bridge)—updating all references across agent definitions, documentation, configuration, and site files. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~30 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Nitpick comments (10)
agents/09-friday.md (1)
160-160: Add language identifiers to fenced code blocks.These fences trigger MD040. Add an explicit language (
textis enough) for consistent lint compliance.💡 Proposed fix
-``` +```text docs/ ├── 00-meta/ # Documentation sur la documentation ... └── 99-archive/-
+text
Mode context :
- Explorer sources (docs, configs, git, audits)
...- Rapport
Also applies to: 248-248
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@agents/09-friday.md` at line 160, The markdown file contains fenced code blocks without language identifiers (triggering MD040); update each fence (e.g., the snippets showing the directory tree and the "Mode context :" list) to include an explicit language tag such as ```text so all fenced code blocks (including those around the docs/ tree and the "Mode context" block referenced near the shown diff and also at the other occurrence around line 248) have a language identifier.agents/05-vision.md (5)
286-305: Add language specifier to startup summary code block.Static analysis flagged this code block as missing a language specifier.
📝 Proposed fix
-``` +```text Mode audit : 1. Reconnaissance (ou skip si contexte reçu)🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@agents/05-vision.md` around lines 286 - 305, The fenced code block under the "## Démarrage" section is missing a language specifier; update the opening fence from ``` to ```text (the block that begins with the line "Mode audit :") so the snippet is tagged as plain text for static analysis and rendering.
148-164: Add language specifier to scoring template code block.Static analysis flagged this code block as missing a language specifier.
📝 Proposed fix
-``` +```text === SCORE D'AUDIT ===🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@agents/05-vision.md` around lines 148 - 164, The fenced code block containing the "=== SCORE D'AUDIT ===" table in agents/05-vision.md is missing a language specifier; update the opening fence for that block (the triple backticks immediately before "=== SCORE D'AUDIT ===") to include a language tag such as text (i.e., change ``` to ```text) so static analysis recognizes it as a text/code block while leaving the table content unchanged.
62-67: Add language specifier to code block.Static analysis flagged this code block as missing a language specifier.
📝 Proposed fix
-``` +=== Cartographie === +```text +=== Cartographie === 📁 Fichiers source : [X] 📊 Lignes de code : [X] 📜 Commits : [X], Contributeurs : [X]🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@agents/05-vision.md` around lines 62 - 67, The fenced code block containing "=== Cartographie ===" is missing a language specifier; update the opening fence from ``` to ```text so the block is explicitly marked as plain text. Locate the block that starts with the line "=== Cartographie ===" and change its opening backticks to include the language token (e.g., ```text) to satisfy the static analysis rule.
228-230: Add language specifier to /simplify command block.Static analysis flagged this code block as missing a language specifier.
📝 Proposed fix
-``` +```text /simplify focus on [issues identifiées]</details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@agents/05-vision.mdaround lines 228 - 230, The fenced code block containing
"/simplify focus on [issues identifiées]" is missing a language specifier;
update the markdown block that wraps that exact text to include a language tag
(use "text") so it becomes a lang-specified fenced block (i.e., change the
openingtotext) to satisfy static analysis and preserve the original
line content.</details> --- `179-181`: **Ensure compliance with update-protocol.md and Monoboard Kanban format.** The agent adds tasks to `docs/todo.md` with prefix "A" (`#A001`, `#A002`), but doesn't explicitly reference the incremental update protocol or Monoboard Kanban format that `shuri` enforces. Per `_shared/update-protocol.md`, vision should: 1. Check if `docs/todo.md` has `kanban-plugin: board` frontmatter 2. Search for existing task IDs before creating new ones 3. Place cards in the correct column (Todo for P0-P2, Backlog for P3) 4. Use the Monoboard card format with zone/effort/dependencies Consider adding an explicit reference to follow `_shared/update-protocol.md` for these writes. <details> <summary>📝 Proposed addition for update protocol reference</summary> ```diff ## Phase A5 : Mise à jour spec/todo -Ajouter section audit dans `docs/spec.md` + nouvelles tâches dans `docs/todo.md` (préfixe `A` pour audit : `#A001`, `#A002`...). +Ajouter section audit dans `docs/spec.md` + nouvelles tâches dans `docs/todo.md` (préfixe `A` pour audit : `#A001`, `#A002`...). + +> **Suivre `_shared/update-protocol.md`** : mise à jour incrémentale, détection du format Monoboard Kanban, placement dans la bonne colonne. ``` </details> Based on learnings: "Documentation update operations must follow the incremental update protocol: use section ownership via emoji headers, check for duplicates before adding tasks, consolidate writes to shared files at single write points". <details> <summary>🤖 Prompt for AI Agents</summary> ``` Verify each finding against the current code and only fix it if needed. In `@agents/05-vision.md` around lines 179 - 181, Update the Phase A5 write steps to explicitly follow the incremental update protocol in _shared/update-protocol.md: before appending to docs/todo.md verify the file has frontmatter key "kanban-plugin: board", scan the file for existing task IDs (e.g., `#A001`, `#A002`) to avoid duplicates, and insert new Monoboard-formatted cards (include zone/effort/dependencies metadata) into the correct column mapping (Todo for P0–P2, Backlog for P3). Consolidate all writes to docs/todo.md into a single write operation and add a short comment/reference in the spec update mentioning adherence to _shared/update-protocol.md and Monoboard Kanban format. ``` </details> </blockquote></details> <details> <summary>agents/CLAUDE.md (1)</summary><blockquote> `307-317`: **Add language specifiers to fenced code blocks.** Static analysis flagged that the code blocks at lines 308 and 320 are missing language specifiers. <details> <summary>📝 Proposed fix for missing language specifiers</summary> ```diff -``` +```text "Import from Notion" → bifrost mode=import (21) → Reads master page + all linked pages recursively → Generates: docs/imports/spec_notion.md + todo_notion.md ``` And similarly for line 320: ```diff -``` +```text "Update local documentation" → shuri mode=sync (01) ``` </details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@agents/CLAUDE.mdaround lines 307 - 317, Add language specifiers to the two
fenced code blocks that lack them: the block containing the "Import from Notion"
snippet (the lines starting with "Import from Notion" → bifrost mode=import) and
the block containing the "Update local documentation" snippet (the line starting
with "Update local documentation" → shuri mode=sync). Change the opening fences
fromtotext (or another appropriate language like ```bash) so the blocks
become fenced with a language identifier and ensure consistency across both
blocks.</details> </blockquote></details> <details> <summary>agents/01-shuri.md (1)</summary><blockquote> `243-249`: **Verify Monoboard format detection handles edge cases.** The logic checks for `kanban-plugin: board` or legacy `## P0`, `## P1` sections, but the condition at line 247 uses `P0/P1...` which could be ambiguous. Consider documenting more explicit detection patterns. <details> <summary>🤖 Prompt for AI Agents</summary> ``` Verify each finding against the current code and only fix it if needed. In `@agents/01-shuri.md` around lines 243 - 249, The Monoboard format detection is ambiguous; update the text and detection logic to use explicit patterns: check for the exact YAML line "kanban-plugin: board" using a strict regex (e.g. ^kanban-plugin:\s*board\b) and detect legacy sections with explicit heading patterns like "^##\s*P\d+\b" (for P0, P1, etc.), then document these patterns in the "Mode mise à jour incrémentale" block so readers know the exact matching rules and edge cases the code uses. ``` </details> </blockquote></details> <details> <summary>agents/Readme.md (2)</summary><blockquote> `65-82`: **Consider adding language specifier to fenced code block.** The code block at line 65 lacks a language identifier, which triggers a markdownlint warning. Since this shows a directory tree structure, you could use `text` or leave it unspecified if intentional. <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@agents/Readme.mdaround lines 65 - 82, The fenced code block that shows the
agents/ directory tree is missing a language specifier; update the block
delimiter (the triple backticks surrounding the agents/ tree) to include a
language tag such as text (e.g., ```text) so markdownlint stops warning, i.e.
locate the fenced block containing the agents/ directory listing and change its
opening fence to include the language specifier.</details> --- `406-408`: **Bifrost mode descriptions use inconsistent capitalization.** The bullet points mix sentence fragments with capitalized terms. Minor inconsistency but the information is accurate. ```diff ### 21-bifrost (Notion Bridge) -- mode=import : Import unidirectionnel Notion → Local (docs/imports/) -- mode=export : Markdown → Notion QA avec User Stories +- `mode=import` : import unidirectionnel Notion → local (docs/imports/) +- `mode=export` : Markdown → Notion QA avec User Stories ``` <details> <summary>🤖 Prompt for AI Agents</summary> ``` Verify each finding against the current code and only fix it if needed. In `@agents/Readme.md` around lines 406 - 408, The two bullet lines under the "21-bifrost (Notion Bridge)" section use inconsistent capitalization for the mode descriptions; pick a consistent style (e.g., sentence case) and apply it to both entries: update the "mode=import : Import unidirectionnel Notion → Local (docs/imports/)" and "mode=export : Markdown → Notion QA avec User Stories" lines so they follow the same capitalization convention (for example: "mode=import : Import unidirectional Notion → local (docs/imports/)" and "mode=export : Markdown → Notion QA with user stories" or adjust to title case consistently), ensuring both "mode=import" and "mode=export" descriptions match. ``` </details> </blockquote></details> </blockquote></details> <details> <summary>🤖 Prompt for all review comments with AI agents</summary>Verify each finding against the current code and only fix it if needed.
Inline comments:
In@agents/09-friday.md:
- Line 128: The migration step currently instructs deleting llm.txt which
conflicts with the non-destructive policy ("Ne jamais supprimer de contenu ;
archiver."); update the migration behavior so that if llm.txt exists it is
moved/archived rather than deleted — for example, move its contents into
docs/context.md and rename or copy the original to a traceable archive (e.g.,
llm.txt.migrated or an archive folder) and update the Migration note to reflect
the non-destructive flow so all references to removing llm.txt are replaced with
the archive/move approach to preserve traceability.In
@agents/21-bifrost.md:
- Line 167: The "MODE EXPORT (Local → Notion)" header is inconsistent with the
documented behavior of thebifrost exportcommand (which is described as
Notion → Markdown); update the header text and any surrounding description to
"MODE EXPORT (Notion → Markdown)" and ensure thebifrost exportsection
content (and any other mentions) consistently describe export direction as
Notion → Markdown so the pipeline behavior matches the documentation.- Around line 149-151: The metadata file path is inconsistent: one place uses
"docs/imports/.notion-import-meta.json" and another uses
".notion-import-meta.json"; pick the canonical path
"docs/imports/.notion-import-meta.json" and update all references to that single
path (e.g., where the file string is used in import/save/load logic and any
checks) so code always reads/writes the same file; search for occurrences of
".notion-import-meta.json" and replace them with
"docs/imports/.notion-import-meta.json" in functions or constants that reference
the path and update any tests or docs that assume the old relative location.In
@commands/README.md:
- Around line 109-122: Update the mode definitions in commands/README.md so they
match the agent names and sequences used in agents/18-blackemperor.md: for
mode=audit replace "shuri spec → [vision + perf-auditor + a11y-auditor] → shuri
todo" with "spec-writer → [code-auditor + perf-auditor + a11y-auditor] →
todo-generator"; for mode=legacy revert "shuri spec → vision audit → [vision
simplify + perf-auditor] → robocop → shuri sync" to "spec-writer → code-auditor
→ [/simplify + perf-auditor]"; and for mode=ship change "vision simplify → shuri
sync → friday context → release" to "/simplify → spec-writer + todo-generator +
CHANGELOG → documentalist → brigitte → sync-local" so the README reflects the
canonical agent names and sequences.
Nitpick comments:
In@agents/01-shuri.md:
- Around line 243-249: The Monoboard format detection is ambiguous; update the
text and detection logic to use explicit patterns: check for the exact YAML line
"kanban-plugin: board" using a strict regex (e.g. ^kanban-plugin:\sboard\b) and
detect legacy sections with explicit heading patterns like "^##\sP\d+\b" (for
P0, P1, etc.), then document these patterns in the "Mode mise à jour
incrémentale" block so readers know the exact matching rules and edge cases the
code uses.In
@agents/05-vision.md:
- Around line 286-305: The fenced code block under the "## Démarrage" section is
missing a language specifier; update the opening fence fromtotext (the
block that begins with the line "Mode audit :") so the snippet is tagged as
plain text for static analysis and rendering.- Around line 148-164: The fenced code block containing the "=== SCORE D'AUDIT
===" table in agents/05-vision.md is missing a language specifier; update the
opening fence for that block (the triple backticks immediately before "=== SCORE
D'AUDIT ===") to include a language tag such as text (i.e., change ``` totable content unchanged. - Around line 62-67: The fenced code block containing "=== Cartographie ===" is missing a language specifier; update the opening fence from ``` to ```text so the block is explicitly marked as plain text. Locate the block that starts with the line "=== Cartographie ===" and change its opening backticks to include the language token (e.g., ```text) to satisfy the static analysis rule. - Around line 228-230: The fenced code block containing "/simplify focus on [issues identifiées]" is missing a language specifier; update the markdown block that wraps that exact text to include a language tag (use "text") so it becomes a lang-specified fenced block (i.e., change the opening ``` to ```text) to satisfy static analysis and preserve the original line content. - Around line 179-181: Update the Phase A5 write steps to explicitly follow the incremental update protocol in _shared/update-protocol.md: before appending to docs/todo.md verify the file has frontmatter key "kanban-plugin: board", scan the file for existing task IDs (e.g., `#A001`, `#A002`) to avoid duplicates, and insert new Monoboard-formatted cards (include zone/effort/dependencies metadata) into the correct column mapping (Todo for P0–P2, Backlog for P3). Consolidate all writes to docs/todo.md into a single write operation and add a short comment/reference in the spec update mentioning adherence to _shared/update-protocol.md and Monoboard Kanban format. In `@agents/09-friday.md`: - Line 160: The markdown file contains fenced code blocks without language identifiers (triggering MD040); update each fence (e.g., the snippets showing the directory tree and the "Mode context :" list) to include an explicit language tag such as ```text so all fenced code blocks (including those around the docs/ tree and the "Mode context" block referenced near the shown diff and also at the other occurrence around line 248) have a language identifier. In `@agents/CLAUDE.md`: - Around line 307-317: Add language specifiers to the two fenced code blocks that lack them: the block containing the "Import from Notion" snippet (the lines starting with "Import from Notion" → bifrost mode=import) and the block containing the "Update local documentation" snippet (the line starting with "Update local documentation" → shuri mode=sync). Change the opening fences from ``` to ```text (or another appropriate language like ```bash) so the blocks become fenced with a language identifier and ensure consistency across both blocks. In `@agents/Readme.md`: - Around line 65-82: The fenced code block that shows the agents/ directory tree is missing a language specifier; update the block delimiter (the triple backticks surrounding the agents/ tree) to include a language tag such as text (e.g., ```text) so markdownlint stops warning, i.e. locate the fenced block containing the agents/ directory listing and change its opening fence to include the language specifier. - Around line 406-408: The two bullet lines under the "21-bifrost (Notion Bridge)" section use inconsistent capitalization for the mode descriptions; pick a consistent style (e.g., sentence case) and apply it to both entries: update the "mode=import : Import unidirectionnel Notion → Local (docs/imports/)" and "mode=export : Markdown → Notion QA avec User Stories" lines so they follow the same capitalization convention (for example: "mode=import : Import unidirectional Notion → local (docs/imports/)" and "mode=export : Markdown → Notion QA with user stories" or adjust to title case consistently), ensuring both "mode=import" and "mode=export" descriptions match.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID:
2c8ec74c-d5a8-4659-a6d1-e249695948ec📒 Files selected for processing (49)
.claude/rules/native-features.mdCLAUDE.mdREADME.mdagents/01-shuri.mdagents/01-spec-writer.mdagents/02-todo-generator.mdagents/03-sync-local.mdagents/05-code-auditor.mdagents/05-vision.mdagents/09-context-generator.mdagents/09-friday.mdagents/13-documentalist.mdagents/17-code-simplifier.mdagents/21-bifrost.mdagents/21-notion-importer.mdagents/33-kanban-converter.mdagents/37-beetlejuice.mdagents/CLAUDE.mdagents/Readme.mdcommands/README.mdcommands/agents/01-shuri.mdcommands/agents/01-spec-writer.mdcommands/agents/02-todo-generator.mdcommands/agents/03-sync-local.mdcommands/agents/05-code-auditor.mdcommands/agents/05-vision.mdcommands/agents/09-context-generator.mdcommands/agents/09-friday.mdcommands/agents/13-documentalist.mdcommands/agents/17-code-simplifier.mdcommands/agents/21-bifrost.mdcommands/agents/21-notion-importer.mdcommands/agents/37-beetlejuice.mdcommands/agents/beetlejuice.mdcommands/agents/bifrost.mdcommands/agents/code-auditor.mdcommands/agents/code-simplifier.mdcommands/agents/context-generator.mdcommands/agents/documentalist.mdcommands/agents/friday.mdcommands/agents/notion-importer.mdcommands/agents/shuri.mdcommands/agents/spec-writer.mdcommands/agents/sync-local.mdcommands/agents/todo-generator.mdcommands/agents/vision.mdinstall.shsite/data/commands.jsonsite/index.html💤 Files with no reviewable changes (28)
- agents/09-context-generator.md
- commands/agents/13-documentalist.md
- commands/agents/notion-importer.md
- commands/agents/beetlejuice.md
- commands/agents/code-auditor.md
- commands/agents/17-code-simplifier.md
- commands/agents/21-notion-importer.md
- commands/agents/02-todo-generator.md
- commands/agents/05-code-auditor.md
- commands/agents/37-beetlejuice.md
- commands/agents/03-sync-local.md
- commands/agents/code-simplifier.md
- agents/17-code-simplifier.md
- commands/agents/01-spec-writer.md
- commands/agents/spec-writer.md
- agents/05-code-auditor.md
- commands/agents/09-context-generator.md
- agents/01-spec-writer.md
- agents/33-kanban-converter.md
- agents/37-beetlejuice.md
- agents/21-notion-importer.md
- commands/agents/todo-generator.md
- agents/02-todo-generator.md
- commands/agents/documentalist.md
- agents/13-documentalist.md
- commands/agents/context-generator.md
- commands/agents/sync-local.md
- agents/03-sync-local.md
| 2. **Réduire** : Arborescence, Historique (5 commits) | ||
| 3. **Supprimer** : Audits anciens, Config MCP détaillée | ||
|
|
||
| > **⚠️ Migration** : Si un `llm.txt` existe à la racine, le migrer vers `docs/context.md` et supprimer l'ancien. |
There was a problem hiding this comment.
Resolve the destructive-policy contradiction.
Line 128 instructs deleting llm.txt, while Line 237 states “Ne jamais supprimer de contenu ; archiver.” Keep a single non-destructive behavior (archive/move with traceability) to avoid unsafe execution paths.
💡 Proposed fix
-> **⚠️ Migration** : Si un `llm.txt` existe à la racine, le migrer vers `docs/context.md` et supprimer l'ancien.
+> **⚠️ Migration** : Si un `llm.txt` existe à la racine, le migrer vers `docs/context.md` puis archiver l'ancien (ex: `docs/imports/llm-archive-YYYY-MM-DD.txt`).Based on learnings "Documentation update operations must follow the incremental update protocol" and "Separation of responsibilities... updates local files only / non-destructive flows".
Also applies to: 237-237
🧰 Tools
🪛 LanguageTool
[grammar] ~128-~128: Il y a peut-être une erreur ici
Context: ...dits anciens, Config MCP détaillée > llm.txt existe à ...
(QB_NEW_FR)
[grammar] ~128-~128: Essayez d’enlever « ** »
Context: ...s anciens, Config MCP détaillée > llm.txt existe à la racine, le mig...
(QB_NEW_FR_OTHER_ERROR_IDS_UNNECESSARY_ADJECTIVE)
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@agents/09-friday.md` at line 128, The migration step currently instructs
deleting llm.txt which conflicts with the non-destructive policy ("Ne jamais
supprimer de contenu ; archiver."); update the migration behavior so that if
llm.txt exists it is moved/archived rather than deleted — for example, move its
contents into docs/context.md and rename or copy the original to a traceable
archive (e.g., llm.txt.migrated or an archive folder) and update the Migration
note to reflect the non-destructive flow so all references to removing llm.txt
are replaced with the archive/move approach to preserve traceability.
Comment on lines
+149
to
+151
| ### docs/imports/.notion-import-meta.json | ||
|
|
||
| Métadonnées d'import pour tracking (dernière date, pages, stats). |
There was a problem hiding this comment.
Unify persisted metadata file path.
You define docs/imports/.notion-import-meta.json at Line 149, but Line 301 references .notion-import-meta.json without the directory. Keep one canonical location to prevent split state and duplicate imports.
💡 Proposed fix
-7. **État persisté** — Sauvegarder dans `.beetlejuice-state.json` / `.notion-import-meta.json`
+7. **État persisté** — Sauvegarder dans `docs/imports/.beetlejuice-state.json` / `docs/imports/.notion-import-meta.json`Based on learnings "Notion import metadata must be tracked in .notion-import-meta.json with timestamps and page mappings to enable incremental updates and prevent duplicate imports".
Also applies to: 301-301
🧰 Tools
🪛 LanguageTool
[typographical] ~149-~149: Il manque une espace après le point.
Context: ...ocal ↔ Notion ID) ``` ### docs/imports/.notion-import-meta.json Métadonnées d'import ...
(ESPACE_APRES_POINT)
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@agents/21-bifrost.md` around lines 149 - 151, The metadata file path is
inconsistent: one place uses "docs/imports/.notion-import-meta.json" and another
uses ".notion-import-meta.json"; pick the canonical path
"docs/imports/.notion-import-meta.json" and update all references to that single
path (e.g., where the file string is used in import/save/load logic and any
checks) so code always reads/writes the same file; search for occurrences of
".notion-import-meta.json" and replace them with
"docs/imports/.notion-import-meta.json" in functions or constants that reference
the path and update any tests or docs that assume the old relative location.
|
|
||
| --- | ||
|
|
||
| # MODE EXPORT (Local → Notion) |
There was a problem hiding this comment.
Fix export direction inconsistency.
Line 167 defines export as Local → Notion, but Line 274 documents bifrost export as Notion → Markdown. This ambiguity can trigger the wrong pipeline.
💡 Proposed fix
-| `bifrost export` | Export Notion → Markdown |
+| `bifrost export` | Export Local → Notion |Also applies to: 274-274
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@agents/21-bifrost.md` at line 167, The "MODE EXPORT (Local → Notion)" header
is inconsistent with the documented behavior of the `bifrost export` command
(which is described as Notion → Markdown); update the header text and any
surrounding description to "MODE EXPORT (Notion → Markdown)" and ensure the
`bifrost export` section content (and any other mentions) consistently describe
export direction as Notion → Markdown so the pipeline behavior matches the
documentation.
Comment on lines
+109
to
+122
| ### mode=audit (`"audit-complet"`) | ||
| shuri spec → [vision + perf-auditor + a11y-auditor] (parallèle) → shuri todo | ||
|
|
||
| ### 🔄 mode=legacy (`"legacy-revival"`) | ||
| spec-writer → code-auditor → [code-simplifier + perf-auditor] (parallèle) → robocop → sync-local | ||
| ### mode=legacy (`"legacy-revival"`) | ||
| shuri spec → vision audit → [vision simplify + perf-auditor] (parallèle) → robocop → shuri sync | ||
|
|
||
| ### 🚀 mode=release (`"pre-release"`) | ||
| [code-auditor + perf-auditor + a11y-auditor] (parallèle) → robocop → tests → GO/NO-GO | ||
| ### mode=release (`"pre-release"`) | ||
| [vision + perf-auditor + a11y-auditor] (parallèle) → robocop → tests → GO/NO-GO | ||
|
|
||
| ### 🔍 mode=review (`"review"`) | ||
| ### mode=review (`"review"`) | ||
| extraction exigences → analyse complétude → matrice → gaps → todo | ||
|
|
||
| ### 📦 mode=ship (`"ship"`) | ||
| code-simplifier → sync-local → context-generator → release | ||
| ### mode=ship (`"ship"`) | ||
| vision simplify → shuri sync → friday context → release |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Description: Verify blackemperor.md workflow definitions match this documentation
# Expected: blackemperor.md should use shuri/vision/friday in mode definitions
echo "=== blackemperor.md Mode: AUDIT section ==="
rg -A 10 'Mode: AUDIT' agents/18-blackemperor.md 2>/dev/null | head -15
echo ""
echo "=== blackemperor.md Mode: LEGACY section ==="
rg -A 10 'Mode: LEGACY' agents/18-blackemperor.md 2>/dev/null | head -15
echo ""
echo "=== blackemperor.md Mode: SHIP section ==="
rg -A 10 'Mode: SHIP' agents/18-blackemperor.md 2>/dev/null | head -15Repository: izo/Ulk
Length of output: 1465
Documentation-implementation mismatch for blackemperor modes.
The commands/README.md documentation uses new agent names (shuri, vision, friday):
mode=audit:shuri spec → [vision + perf-auditor + a11y-auditor] → shuri todomode=legacy:shuri spec → vision audit → [vision simplify + perf-auditor] → robocop → shuri syncmode=ship:vision simplify → shuri sync → friday context → release
However, agents/18-blackemperor.md still uses the original agent names:
mode=audit:spec-writer → [code-auditor + perf-auditor + a11y-auditor] → todo-generatormode=legacy:spec-writer → code-auditor → [/simplify + perf-auditor]mode=ship:/simplify → spec-writer + todo-generator + CHANGELOG → documentalist → brigitte → sync-local
Update commands/README.md to match the current agent definitions in agents/18-blackemperor.md.
🧰 Tools
🪛 LanguageTool
[typographical] ~109-~109: Caractère d’apostrophe incorrect.
Context: ...s commandes raccourcies ### mode=audit ("audit-complet") shuri spec → [vision +...
(APOS_INCORRECT)
[typographical] ~112-~112: Caractère d’apostrophe incorrect.
Context: ...arallèle) → shuri todo ### mode=legacy ("legacy-revival") shuri spec → vision a...
(APOS_INCORRECT)
[grammar] ~113-~113: Il y a peut-être une erreur ici
Context: ...dit → [vision simplify + perf-auditor] (parallèle) → robocop → shuri sync ### mode=releas...
(QB_NEW_FR)
[grammar] ~113-~113: Il y a peut-être une erreur ici
Context: ...on simplify + perf-auditor] (parallèle) → robocop → shuri sync ### mode=release (`"pre-rel...
(QB_NEW_FR)
[typographical] ~115-~115: Caractère d’apostrophe incorrect.
Context: ... robocop → shuri sync ### mode=release ("pre-release") [vision + perf-auditor +...
(APOS_INCORRECT)
[typographical] ~118-~118: Caractère d’apostrophe incorrect.
Context: ...cop → tests → GO/NO-GO ### mode=review ("review") extraction exigences → analys...
(APOS_INCORRECT)
[typographical] ~121-~121: Caractère d’apostrophe incorrect.
Context: ... → matrice → gaps → todo ### mode=ship ("ship") vision simplify → shuri sync → ...
(APOS_INCORRECT)
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@commands/README.md` around lines 109 - 122, Update the mode definitions in
commands/README.md so they match the agent names and sequences used in
agents/18-blackemperor.md: for mode=audit replace "shuri spec → [vision +
perf-auditor + a11y-auditor] → shuri todo" with "spec-writer → [code-auditor +
perf-auditor + a11y-auditor] → todo-generator"; for mode=legacy revert "shuri
spec → vision audit → [vision simplify + perf-auditor] → robocop → shuri sync"
to "spec-writer → code-auditor → [/simplify + perf-auditor]"; and for mode=ship
change "vision simplify → shuri sync → friday context → release" to "/simplify →
spec-writer + todo-generator + CHANGELOG → documentalist → brigitte →
sync-local" so the README reflects the canonical agent names and sequences.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This PR consolidates 10 specialized agents into 4 unified pipeline agents, reducing complexity while maintaining all functionality. Each new agent orchestrates related workflows under a single command interface.
Key Changes
New Unified Agents:
Removed Agents:
Updated References:
Implementation Details
Each unified agent maintains:
The consolidation reduces cognitive load for users while preserving granular control through mode selection.
https://claude.ai/code/session_01Sz6RTvd8y7ccHjuVsAmDsp
Summary by CodeRabbit
Release Notes
New Features
Refactor