docs: deepen managed-agent and skills/tools surface documentation#126
Merged
Conversation
Document the CMA agent definition (model, betas, code-enforced A4 guarantee, frontmatter validation, tools/mcp_servers omission) and a reproducible stand-up walkthrough in packages/limner-agent/README.md, correcting the stale "six wave-1 skills" count to the real 16 and expanding the anti-drift contract to the full 19-entry guidance-to- surface matrix. Add Managed Agent Setup and Skills and Capability Surfaces sections to the root README, and Prompts (12) / Resources (3) sections to packages/limner-mcp/README.md alongside a tool-count heading fix (17 -> 18, the table itself was already complete). Corrects an inaccurate "bare tool names" claim repeated in both the mcp README and prior guidance: @limner/cma-tools ships the identical limner_-prefixed tool names as @limner/mcp, in lockstep; bare-name exposure is the CMA platform's own dispatcher behavior, and normalizeToolName() is a Test 4 scoring shim, not a runtime rename. Asana: 1216368512919579 Signed-off-by: Jim Vinson <jim@vinson.org>
vinsonconsulting
added a commit
that referenced
this pull request
Jul 8, 2026
Follow-up to #126, which deepened the managed-agent and skills/tools docs across three READMEs but left this fourth file out of scope. - Corrects the stale "15-tool" count in two spots (intro sentence and the deploy-step note); every other count in the file already said 18, matching the source (LIMNER_TOOLS, 18 entries). - Corrects an inaccurate "exposed under bare names" claim: this package ships the identical limner_-prefixed tool names as @limner/mcp, in lockstep. Bare-name exposure is the Claude Managed Agents platform's own dispatcher behavior, not something this package does (same correction #126 made in the other two READMEs). - Adds the missing limner_create_project row to the tool table, which is why the table only showed 17 named tools despite the prose already saying 18. Asana: 1216368512919579 Signed-off-by: Jim Vinson <jim@vinson.org>
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
1 participant
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
Docs-only PR deepening two under-documented areas: how the CMA agent is
defined and stood up (Path A), and the full capability surface (agent
Skills, MCP Prompts, MCP Resources) that all derive from one guidance core
(Path B). No source, schema, or tool behavior changes.
Per-file changes
packages/limner-agent/README.md— corrected the stale "six wave-1skills" claim (it appeared in two places;
SKILL_MANIFESTactually has 16:6 wave-1 + 10 wave-2/D-RA-24). Added a 16-skill table in attach order with
surface annotations, a "Definition reference" subsection (model, beta
headers, the code-enforced
assertA4()guarantee, frontmatter validationrules, the deliberate tools/mcp_servers omission), a 6-step reproducible
stand-up walkthrough grounded in
create-agent.ts, and expanded theanti-drift contract with the full 19-entry guidance-to-surface matrix.
README.md(root) — added Managed Agent Setup and Skills andCapability Surfaces sections between Memory and Security (plus Contents
entries), with a compact stand-up walkthrough and links down to the full
limner-agentreference.packages/limner-mcp/README.md— fixed the## Tool surface (17)heading to
(18)(the table itself already listed all 18 tools correctly;only the heading was wrong). Added Prompts (12) and Resources (3)
sections. Also corrected an inaccurate claim in the existing tool-surface
paragraph:
@limner/cma-toolsdoes not expose tools under bare names;it ships the identical
limner_-prefixed names as@limner/mcpinlockstep. Bare-name exposure is the CMA platform's own dispatcher behavior;
normalizeToolName()is a Test 4 scoring shim, not a runtime rename.CHANGELOG.md— added an### Addedentry under[Unreleased]matching the existing style.
Reconciled counts
16 skills · 12 prompts · 3 resources · 18 tools (19 guidance entries render
into the first three;
style-profilebacks no standalone surface).Verification
pnpm build(root): passpnpm lint(root): pre-existing failures only, in gitignored local scratchpaths (
.remember/tmp/,test-runs/) untouched by this diff and notpresent in
git statusbefore or after; this diff touches Markdown only,which ESLint doesn't lint
pnpm --filter @limner/limner-agent test: 98/98 pass, includingskills-drift.test.ts(generatedSKILL.mdregions untouched)limner-mcp/README.mdleft as-is, out of scope)A4 framing
Held throughout; no claim of Anthropic/Cloudflare endorsement introduced.
Asana: 1216368512919579
🤖 Generated with Claude Code