agentleFS
preetamnathpreetamnath/agent-skills/skills/place-fact/SKILL.md

place-fact

Skill4 starsChanged 7 months agoMITView on GitHub

Route guidance, state, or evidence already worth keeping to one canonical owner and the narrowest reliable delivery boundary. TRIGGER when: user asks 'where should this go', 'which file/rule/home does this belong in', 'is this in the right place', or 'should this be a pointer'; placing or re-homing kept material.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    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
  1. Distinctive:

    Create: Create an artifact when its trigger and scope form a distinct delivery boundary.

    line 27
  2. Distinctive:

    Judge bloat: Look for duplicated ownership, mixed scopes, broken routes, and unnecessary eager loading.

    line 30
  3. Distinctive:

    Delivery: Keep one write path; use a relative symlink, native import, or generated view for another loader.

    line 33
  4. Distinctive:

    Prefer executable guards: Keep prose only for a decision, rationale, or required companion action that detection alone cannot convey.

    line 37
  5. Very specific:

    Store Claude path rules once: Keep each file-matching rule in .claude/rules/, with quoted paths: globs.

    line 40
  6. Distinctive:

    Point to Claude path rules only for gaps: Use one only for a cross-layer obligation or new-file gap.

    line 42
  7. Distinctive:

    Skip redundant routing: Never point one AGENTS.md to another that already auto-loads, and never maintain a folder-to-owner map.

    line 43

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Place Fact

Primitive: PLACE — which durable owner and delivery boundary should hold this material?

Steps

  1. Name the delivery trigger — the moment a future agent must already hold the material. A topic, folder, or existing file does not choose the home; the trigger does.
  2. Map the trigger to the narrowest reliable home:

    TriggerHomeAdmit only when
    Any repository taskRoot AGENTS.mdThe instruction or route is repository-wide and non-derivable.
    Reading or editing one mechanismComment or docstring beside itThe mechanism needs a constraint, assumption, or tempting-wrong-path warning.
    Reading, editing, or creating files in one cohesive subtreeNearest ancestor AGENTS.mdA real module boundary has one rule for every current and future file below it.
    Working with files selected by path patterns rather than directory hierarchyClaude path rule in .claude/rules/Quoted paths: globs scope the instruction to matching files.
    Starting a named product, design, operations, or decision taskMaintained task documentCross-cutting context has an inbound route and a workflow that keeps it current.
    Running a repeatable repository-local procedureNamed workflow documentThe procedure remains multi-step and repository-specific.
    Running a repeatable cross-repository or external-platform procedureSkillThe procedure has a reliable trigger; a repository-internal fact alone is not a skill.
    Tracking queued or in-progress workActive-state documentA workflow reads, updates, and retires it.
    Preserving a scoped question, evidence, provisional findings, or open decisionsDated investigation outside ordinary auto-loadClosure promotes current conclusions, routes unfinished work to active state, and retains only unique evidence.
  3. Choose the delivery boundary before considering file reuse:
    • Create. Create an artifact when its trigger and scope form a distinct delivery boundary.
    • Reuse. Reuse an artifact only when its trigger, scope, audience, upkeep, and retirement path already match.
    • Split. Split an artifact when it mixes material with different triggers or scopes.
    • Judge bloat. Look for duplicated ownership, mixed scopes, broken routes, and unnecessary eager loading; file count alone is not bloat.
  4. Confirm the full placement contract:
    • Owner and scope. Name one canonical owner and the exact paths, subtree, task, or procedure it governs.
    • Delivery. Keep one write path; use a relative symlink, native import, or generated view for another loader. Fall back to a copy only when derivation cannot work, and guard equality automatically.
    • Upkeep. Name what keeps the material current: editing it beside its source, a workflow, or an executable guard.
    • Retirement. Name the event that removes or re-homes the material.
  5. Prove the delivery before committing the placement:
    • Prefer executable guards. Keep prose only for a decision, rationale, or required companion action that detection alone cannot convey.
    • Verify every agent. Confirm every supported agent discovers the material at its trigger and, when applicable, after compaction.
    • Cover future files. A Claude path rule may not fire on a new-file write; put a convention governing future files in the nearest AGENTS.md.
    • Store Claude path rules once. Keep each file-matching rule in .claude/rules/, with quoted paths: globs.
    • Point only when needed. Point from AGENTS.md only to a must-read target that will not already auto-load.
    • Point to Claude path rules only for gaps. Use one only for a cross-layer obligation or new-file gap; otherwise let the rule fire or widen its paths.
    • Skip redundant routing. Never point one AGENTS.md to another that already auto-loads, and never maintain a folder-to-owner map.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.