capability-planning
Skill2 starsChanged 32 days agoMITView on GitHub
Runs the capability-aware planning pipeline and renders IMPLEMENTATION_PLAN sections
In this repository52 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .cursor/rules/00-core-operating-model.mdcCursor rule
- .cursor/rules/01-plan-approval-gate.mdcCursor rule
- .cursor/rules/02-git-worktree-policy.mdcCursor rule
- .cursor/rules/03-validation-definition-of-done.mdcCursor rule
- .cursor/rules/04-documentation-memory.mdcCursor rule
- accessibility-reviewSkill
- agent-skills-packagingSkill
- autonomy-budgetSkill
- bounded-autonomySkill
- candidate-promotionSkill
- code-reviewerSkill
- code-structure-cleanupSkill
- compass-evaluatorSkill
- context-inodesSkill
- dependency-supply-chainSkill
- docker-cloudSkill
- embedding-providersSkill
- execution-telemetrySkill
- experience-routingSkill
- experience-skill-trainingSkill
- external-knowledge-ingestSkill
- github-integrationSkill
- goose-agent-orchestrationSkill
- harness-gcSkill
- hosted-vector-dbSkill
- implementation-planningSkill
- ios-engineeringSkill
- knowledge-stewardSkill
- linear-integrationSkill
- node-engineeringSkill
- northstar-connected-routineSkill
- notion-integrationSkill
- package-registry-tiSkill
- persistent-role-promotionSkill
- playwright-browser-validationSkill
- postgres-prismaSkill
- procedure-playbooksSkill
- project-overseerSkill
- pull-request-preparationSkill
- python-mlSkill
- react-engineeringSkill
- repository-discoverySkill
- review-fix-loopSkill
- security-audit-proceduresSkill
- security-reviewSkill
- skill-learning-loopSkill
- skill-lifecycleSkill
- source-code-contextSkill
- technology-intelligence-liveSkill
- testing-validationSkill
- worktree-orchestrationSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 29
Ensure registry is current: ./scripts/compile-capability-registry.sh
- Distinctive: line 30
Run the planning pipeline:
- Distinctive: line 46
If Capability Gaps are present, surface them explicitly — do not silently improvise.
- Distinctive: line 81
Do not treat Technology Intelligence candidates as approved Skills
- Distinctive: line 82
Do not begin product implementation
- Distinctive: line 83
Do not omit capability gap sections when gaps exist
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Capability-Aware Planning
Use this Skill when
The First Mate is writing or updating IMPLEMENTATION_PLAN.md for a non-trivial
objective and needs machine-assisted:
- required capability inference
- Skill discovery and ranking
- task graph decomposition
- agent manifest proposals
Load this Skill together with implementation-planning during /plan-feature.
Inputs
- User objective / request text
- Repository root
- Optional context: stacks,
PROJECT_CONTEXT.mdexcerpt, discovery summary - Plan ID (defaults to
draftuntil issue assigned)
Procedure
- Ensure registry is current:
./scripts/compile-capability-registry.sh Run the planning pipeline:
./scripts/capability-plan.sh --plan-id <plan-id> "<objective>"- Copy rendered sections into
IMPLEMENTATION_PLAN.md:- Required Capabilities
- Reusable Capabilities Found
- Technology Intelligence Candidates
- Task Graph
- Proposed Agent Configuration
- Evaluation Strategy
- Learning Plan
- Approval Boundary
- Complete human-authored sections (Problem Statement, Acceptance Criteria, risks, etc.) using
implementation-planning. - If Capability Gaps are present, surface them explicitly — do not silently improvise.
- Set plan status to AWAITING APPROVAL and stop.
Machine artifacts
Written under .agent/plans/<plan-id>/:
resolve.jsontask-graph.jsonmanifests.json
M40: typed dependency links + subagent pwd
task-graph.json dependencies carry explicit link types (backward compatible):
- string form (legacy) — pure ordering edge
- hard link —
{ "target", "link": "hard", "contract" }: version-locked contract coupling; both sides must change together (e.g. implementation tasks hard-link thearchitecture-briefcontract) - symlink —
{ "target", "link": "symlink", "paths" }: loose module / artifact path reference (e.g. validation symlinks implementation artifacts)
Manifests include a working_context block (the subagent pwd):
context_root (scoped route in .agent/context/), inode_refs (only the
in-scope module's inode pointers), and scope_allow / scope_deny path lists.
Disjoint module tasks must not share inode refs — that isolation invariant is
checked by the M40 eval sensor. Build the inode store first (Skill
context-inodes) or manifests degrade to unrestricted scope with a note.
Output
Capability-aware plan sections ready for Captain review.
Prohibited actions
- Do not treat Technology Intelligence candidates as approved Skills
- Do not begin product implementation
- Do not omit capability gap sections when gaps exist
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

