agentleFS
Sign inStart free
iwesysiwesys/IWE/.claude/skills/fpf/SKILL.md

fpf

Skill0 starsChanged 25 days agoMITView on GitHub

Load applicable principles for a task from the Pack → SPF → FPF hierarchy.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    BLOCKING — No-Jargon: Every FPF term in output → engineering equivalent in parentheses or in place of the term.

    line 51
  2. Very specific:

    If the FPF/ folder is missing or the required pattern is not in the full text → explicitly state "full text not accessible" and leave …

    line 66
  3. Very specific:

    Search: rg -n -i -C 2 --no-ignore -g '*.md' -- '<term in English and Russian>' FPF.

    line 67
  4. Very specific:

    Read a complete pattern by ID (escape the dot in the ID with \.): rg -U --no-heading --no-line-number --no-filename --color never '(?ms)^## …

    line 68
  5. Distinctive:

    Use the table above. If unclear — use lookup mode.

    line 73
  6. Distinctive:

    If the first level returns no results — follow the fallback chain

    line 86

≈ “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

Loading Principles

Request: $ARGUMENTS

When to use

Load applicable principles for a task from the Pack → SPF → FPF hierarchy. Use when you need to find relevant principles before making a decision.

Fallback Chain

Pack (domain-specific) → SPF (correctness) → FPF (first principles)

Modes (determine from request context)

ModeTriggerAction
lookupRequest contains a pattern code (A.7, B.5, F.18) or a specific termGrep through FPF → exact section. Explain in engineering language (see memory/fpf-reference.md § Translation)
review"Check", "verify against FPF", artifact submitted for reviewLoad artifact + relevant principles → point-by-point verification → findings
design"Design", "how to do this per FPF", design taskSelect applicable patterns (A.1 boundaries, A.6 composition, A.7 distinctions, B.5 ADI) → concrete recommendations
characterize"Compare", "evaluate options", "how to choose"A.17-A.19 (Lawful Comparison): criteria → indicators → evaluate options on a common scale
teach"explain FPF", "what is FPF", "from scratch", "introduction"Give an FPF overview: what it is, why it matters, key blocks (distinctions, patterns, first principles) → explain in engineering language without jargon
diagnose"which principles am I violating", "what is wrong", "find the problem", "review my work"Get a description of the work/solution → find deviations from FPF/Pack principles → name specific violations with the principle and its location
prioritize"most important", "for a role", "in the role", "for me as"Identify the role from the argument → find FPF/Pack principles relevant to that role → rank by applicability → explain why each matters for that role specifically

BLOCKING — No-Jargon: Every FPF term in output → engineering equivalent in parentheses or in place of the term. No exceptions.

  • Known terms: memory/fpf-reference.md § Translation (Holon, F-G-R, Alpha, Mereology, Affordance, ADI...)
  • Unknown terms (TransformerRole, senseFamily, Γ_*, ReferencePlane, etc.): translate by context, add (FPF: <original>). Do not leave any FPF term without translation.
  • Pre-output check: scan the output — is there any term an engineer without FPF background would not understand? If yes — translate it.

Algorithm

0. Verify you are reading the full text, not a reference

A link to FPF, a list of file names, or a truncated fragment is not access to the text (author rule, FPF/USING-FPF.md).

This step applies to the local copy of FPF/ in the workspace. Paths below are relative to its root ($IWE_WORKSPACE); paths with spaces (Engineering DPF Suite/...) must be quoted. In the browser version where no local copy exists, or when search runs via MCP: read the document in full (knowledge_get_document) and name the source. If neither a local copy nor MCP access to the full text is available → state that the full text is unavailable and suggest installing a copy (bash update.sh) or connecting MCP.

  • Confirm that FPF/USING-FPF.md exists and read it if you have not done so in this session: it is the author's general guide for searching and reading patterns.
  • State the copy version: git -C FPF log -1 --format='%h %ad' --date=short. If the commit date is more than 30 days old or USING-FPF.md is missing → tell the user: the copy is outdated; run update.sh to update it (or manually git -C FPF pull --ff-only if the copy has no local edits).
  • If the FPF/ folder is missing or the required pattern is not in the full text → explicitly state "full text not accessible" and leave dependent conclusions open. Do not reconstruct patterns from memory.
  • Search: rg -n -i -C 2 --no-ignore -g '*.md' -- '<term in English and Russian>' FPF. A match in a table of contents or index is not the pattern body.
  • Read a complete pattern by ID (escape the dot in the ID with \.): rg -U --no-heading --no-line-number --no-filename --color never '(?ms)^## <ID> [^\r\n]*\r?\n.*?^### <ID>:End[ \t]*\r?$' -- <file>. If output is truncated → read the same file by line range.
  • No rg (ripgrep) → use the same steps with grep -rn -i --include='*.md' -- '<term>' FPF (recursive, without **, which is not recursive in bash without globstar) and read by line ranges (sed -n 'A,Bp'): start at the line ## <ID> , end at ### <ID>:End.

1. Determine mode

Use the table above. If unclear — use lookup mode.

2. Classify the source

  • Domain knowledge (architecture, domain, service) → source_type=pack
  • Form, process, correctness → source=SPF
  • Base distinctions, first principles → source=FPF

3. Search via iwe-knowledge

  • iwe-knowledge search(query="<query>", source_type="pack") — across all Packs
  • iwe-knowledge search(query="<query>", source="SPF") — across SPF
  • iwe-knowledge search(query="<query>", source="FPF") — across FPF
  • If the first level returns no results — follow the fallback chain

4. If iwe-knowledge is unavailable

(not in /mcp):

  • Pack: read PACK-*/pack/ files via Glob + Read
  • SPF: read SPF/docs/ via Glob + Read
  • FPF: read FPF/Readme.md (overview) or search via Grep across FPF/
  • DPF suites: alongside the specification, the FPF author publishes FPF/Engineering DPF Suite/ (reference: ENGINEERING-DPF-SUITE-REFERENCE.md) and FPF/Foundational Thinking DPF Suite/ (reference: FOUNDATIONAL-THINKING-DPF-SUITE-REFERENCE.md); these are not part of FPF-Spec.md. Entry rules are in FPF/USING-FPF.md: start from a real situation, select a suite via FPF/Readme.md and the suite reference; locate a pattern by its heading ## <ID> - …, its end marker is ### <ID>:End. If these folders are absent from the local FPF/ copy, the copy is outdated — update it.

5. Show the result

List the relevant principles with an explanation of how they apply to the task. Include a reference to the source (github_url from the search result or the file path). Translate FPF terms into engineering language.

Source at the end of the response. State: which publication and copy version (commit), pattern numbers and sections used, what you are assuming, and what information is missing for a complete conclusion. Verbatim quotation is only necessary when verifying an interpretation, not in every response.

<!-- USER-SPACE --><!-- /USER-SPACE -->

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.

Check your setup, free