Agent skill

Ijfw Verify

by FerroxLabs in FerroxLabs/ijfw

A skill your agent uses when about to claim completion: 'done', 'fix complete', 'tests pass', 'build succeeded', 'shipped', 'no regressions', 'ready to merge', 'ready to ship'.

MITAuto-check passedAgent Workflows

Install Ijfw Verify

skills CLI
$ npx skills add FerroxLabs/ijfw --skill ijfw-verify -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install FerroxLabs/ijfw ijfw-verify --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/FerroxLabs/ijfw.git skills-src && mkdir -p .claude/skills && cp -r skills-src/claude/skills/ijfw-verify .claude/skills/ijfw-verify && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
ijfw-verify
GitHub stars
212
Token cost
~2.4k tokens
SKILL.md length
1,261 words
Files
1
Skills in repo
38
Repo updated
First seen
Licence
MIT

At a glance

A skill your agent uses when about to claim completion: 'done', 'fix complete', 'tests pass', 'build succeeded', 'shipped', 'no regressions', 'ready to merge', 'ready to ship'.

  • Works in 5 steps: IDENTIFY -- What command, in this repo,… → RUN -- Execute the FULL command via Bash… → READ -- Read the full output. Check the… → …
  • About to claim completion: done
  • SKILL.md covers The Iron Law, Why this exists, The 5-Step Gate Function and Common Failures -- claim vs…, plus 8 more sections
  • Calls npm, git and curl

What it does

Ijfw Verify is an agent skill from FerroxLabs/ijfw. Use when about to claim completion: 'done', 'fix complete', 'tests pass', 'build succeeded', 'shipped', 'no regressions', 'ready to merge', 'ready to ship'. Iron Law gate requiring fresh verification evidence in the same message as the claim; wires into runtime (verification-gate.js + the ijfwstate MCP tool subagent.post-done verb).

Its SKILL.md is about 2.4k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering MCP servers and Subagents. The repository describes itself as: IJFW — It Just Fcking Works. Ferrox Labs' local-first infrastructure for AI coding agents: shared memory, smart routing, multi-AI cross-audits, disciplined workflow. The licence is MIT.

When your agent uses it

  • About to claim completion: done
  • Build succeeded

Example prompts

  • “fix complete”
  • “tests pass”
  • “build succeeded”
  • “/ijfw-verify”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. IDENTIFY -- What command, in this repo, proves the claim true?
  2. RUN -- Execute the FULL command via Bash tool, in the same message as the claim. Not summarised. Not paraphrased. The actual command, fresh.
  3. READ -- Read the full output. Check the exit code. Count failures. Do not skim.
  4. VERIFY -- Does the output literally confirm the claim?
  5. ONLY THEN -- emit the claim.

What it can do on your machine

Read from SKILL.md and the folder at commit eda62f3. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • npm
    • git
    • curl
    • node

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use npm, git and curl, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Ijfw Verify loads about 2.4k tokens when it runs. Until then it costs about 87 tokens; SKILL.md has 1,261 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~87
When it runs · the whole SKILL.md, loaded when a task matches
~2.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from FerroxLabs/ijfw at commit eda62f3, republished under its MIT licence (© FerroxLabs). 1,261 words, ~2,379 tokens.

Download SKILL.mdSave it as .claude/skills/ijfw-verify/SKILL.md (or your agent's skills folder).
name
ijfw-verify
description
Use when about to claim completion: 'done', 'fix complete', 'tests pass', 'build succeeded', 'shipped', 'no regressions', 'ready to merge', 'ready to ship'. Iron Law gate requiring fresh verification evidence in the same message as the claim; wires into runtime (verification-gate.js + the ijfw_state MCP tool subagent.post-done verb).

IJFW Verify -- The Iron Law

The Iron Law

NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE.

If you haven't run the verification command in this message, you cannot claim it passes. "Earlier" doesn't count. "Should pass" doesn't count. "The code looks right" doesn't count.

Violating the letter of this rule is violating the spirit of this rule.

Why this exists

IJFW v1.4.x shipped at least four times with "done" claims that later turned out to be false: a v1.4.0 milestone declared "shipped" without verifying the GitHub mirror push had landed; v1.4.3 declared Windows CI "promoted to required" before the workflow file change had been merged; v1.4.4 had three of six parallel subagents return DONE without their checkpoints being written; and the Trident cross-audit at r13 reported PASS while one auditor (codex) was UNREACHABLE. Every one of those was an Iron-Law violation -- a claim without same-message evidence -- and every one cost a recovery wave.

This skill is the gate that catches it.

The 5-Step Gate Function

Before emitting any completion claim (verbal or structured):

  1. IDENTIFY -- What command, in this repo, proves the claim true?
  2. RUN -- Execute the FULL command via Bash tool, in the same message as the claim. Not summarised. Not paraphrased. The actual command, fresh.
  3. READ -- Read the full output. Check the exit code. Count failures. Do not skim.
  4. VERIFY -- Does the output literally confirm the claim?
    • If NO: state the actual status with the evidence. Do not claim completion.
    • If YES: state the claim with the output excerpted inline.
  5. ONLY THEN -- emit the claim.

Skip any step = lying, not verifying.

Common Failures -- claim vs required evidence

ClaimRequired evidence (same message)
"all tests pass"Output of npm test (or the project's test command), exit 0, failure count = 0
"build succeeded"Output of the build command, exit 0
"fixed the bug"Reproduction output before + after the fix
"deployment complete"Deployed URL + health-check (curl -fsS <url>) output
"implementation done"Output proving the new behaviour (run the new code)
"no regressions"Baseline test output + post-change test output
"all subagents DONE"cat .ijfw/wave-*/subagent-*.checkpoint.json showing each checkpoint written
"wave complete"node scripts/wave-status.js showing all subs DONE and commits resolved
"Trident r<N> PASS"All three auditor reports written and readable; no UNREACHABLE
"shipped to npm"npm view <pkg> version returning the new version
"pushed to GitLab/GitHub"git ls-remote <remote> <tag> returning the tag SHA
"ledger clean"cat .ijfw/state/execute-issues.json showing zero unresolved entries

If your claim is not in this table, the table is not exhaustive -- derive the required evidence from the spirit of the rule (see "Spirit over letter" below).

Rationalization Prevention -- the lies you tell yourself

RationalizationCounter
"I'm in a hurry"The hurry is exactly when evidence matters most. Hurry shipped v1.4.0 with a missing mirror.
"The code looks right"Code reading proves nothing. Running it does.
"Tests should pass"Run them. "Should" is not evidence.
"I already verified earlier"Verify in the SAME message as the claim. State drifts; context windows lie.
"The subagent reported DONE"Subagent reports are unverified by definition. Check the checkpoint file and the commit SHA.
"Trust the workflow state".ijfw/state/workflow.json records what was claimed, not what happened. Verify the artefact.
"I'm confident"Confidence is not evidence. Three v1.4.x ships had confident authors.
"It worked last time"Last time is not this time. Run it.
"The diff looks plausible""Plausibility is not correctness." (Verify-command preamble exists because people were shipping plausible diffs.)
"I'm tired"Exhaustion is not an excuse. Stop and rest, or run the command.
"Just this once"There is no "just this once". Every Iron-Law violation in IJFW history was a "just this once".
"Different words so the rule doesn't apply"Spirit over letter. Paraphrased success is still a success claim.

Red flags -- STOP and run the gate

  • About to type "should", "probably", "looks like", "seems to"
  • About to type "Great!", "Perfect!", "Done!", "Shipped!"
  • About to commit, push, tag, or open a PR
  • About to mark a workflow phase complete
  • About to emit Status: DONE in a subagent handoff
  • About to trust a subagent / auditor / hook success report without re-checking the artefact
Show full SKILL.md (580 more words)Show less

Integration with the runtime gate

This skill is the author-time half of IJFW's verification contract. The runtime half lives in mcp-server/src/orchestrator/verification-gate.js::checkVerificationGate (originally shipped in v1.4.4 as N5 as an advisory check) and is now invoked automatically by mcp-server/src/orchestrator/post-done-runner.js whenever the ijfw_state MCP tool's subagent.post-done verb fires after a subagent emits Status: DONE (v1.5.0-major / v1.5.0 T13 — single state-SDK MCP face).

What that means in practice:

  • If you skip the Iron Law and emit Status: DONE anyway, the post-done runner will run checkVerificationGate against your checkpoint + commit + test output and downgrade the status to DONE_WITH_CONCERNS (or BLOCKED) when it detects an unsupported claim.
  • The runtime gate is the safety net. It is not the gate. This skill is the gate. The runtime gate exists because skills are advisory and humans (and LLMs) sometimes lie; do not treat the runtime check as permission to skip the author-time check.
  • If you find yourself thinking "the runtime gate will catch it" -- that thought is the Iron-Law violation. Run the command.

Existing IJFW Verify-phase ledger gate (still applies)

Before emitting VERIFY PASS for a workflow phase, the existing ledger check still runs:

bash
read_issues() {
  local f=".ijfw/state/execute-issues.json"
  [ -f "$f" ] || { printf '{"issues":[]}'; return; }
  cat "$f"
}

Any entry with status: unresolved (task-incomplete, task-stagnated, unsafe-verify, plan-review) blocks VERIFY PASS. Missing file = zero issues (day-1 fresh-install protection). This skill adds the Iron Law on top; it does not replace the ledger gate.

Confidence declaration (still required at end of Verify)

Every Verify finding is tagged:

  • VERIFIED -- command was run in this message, raw output is excerpted, anyone can reproduce.
  • LIKELY -- reasoning provided (code read, docs consulted), not externally verified this session. Requires user acknowledgement to advance.
  • GUESSING -- insufficient information; best guess only. Blocks advance.
  • ISSUE -- blocker or bug surfaced; halt and document.

Only VERIFIED findings clear the ship gate unattended. The Iron Law is what earns the VERIFIED tag -- a finding is not VERIFIED unless the verification command + output appears in the same message.

Spirit over letter -- anti-loophole clause

Evidence patterns in the tables above are examples, not exhaustive. The rule is:

Every completion claim in this message must be falsifiable from the tool calls in this message.

If a clever workaround would still leave a future reader unable to independently verify the claim from the message contents, the workaround is not allowed. Examples of disallowed workarounds:

  • Citing an exit code without showing the command that produced it.
  • Pointing at a previous message's output ("see above"); copy it inline or re-run.
  • Trusting an agent/auditor/hook report without checking the underlying artefact (commit SHA, file contents, npm registry, remote tag).
  • Asserting "tests pass" because the test file exists and was edited.
  • Asserting "shipped" because a tag was created locally without git push verification.
  • Asserting "Trident PASS" because the synthesis file exists, without checking each auditor's input file is non-empty and reports PASS.

Iron-Law discipline -- the four boundaries

The runtime gate fires at four enumerated W3 boundaries -- the verbs that advance state ABOVE you:

  • phase.complete -- post-phase verdict; enforceVerificationGate refuses on red.
  • phase.plan-check -- pre-execute plan validation; any HIGH-tier finding refuses dispatch.
  • subagent.post-done -- subagent completion; runSelfCheck refuses when claimed files/commits aren't on disk.
  • wave.advance (hard gate) -- mid-wave checkpoint-completeness; a wave that declares hard_gate: true refuses to advance while any registered subagent lacks a checkpoint.

Discipline rule: collect verification evidence after the change, in the same message as the claim, before the boundary verb fires. The gate is the floor, not the ceiling -- pre-empting it is the job.

The bottom line

No shortcuts for verification.

Run the command. Read the output. Then claim the result.

This is non-negotiable.

© FerroxLabs, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in claude/skills/ijfw-verify of FerroxLabs/ijfw.

Open the folder on GitHubat commit eda62f3

Compare with similar skills

Ijfw Verify next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Ijfw Verify compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Ijfw Verify this skillFerroxLabs/ijfw212—~2.4kAutomated safety check: PassMIT
Claude Automation Recommenderanthropics/claude-plugins-official38k3 repos~2.7kAutomated safety check: NotesApache-2.0
CC Workflow Studio AI Editorbreaking-brake/cc-wf-studio5.4k—~561Automated safety check: PassCustom licence
Agent Deckasheshgoplani/agent-deck1k—~1.7kAutomated safety check: PassMIT
Bootstrapinfragate/capa724—~5.9kAutomated safety check: PassMIT
Puppetmaster Agent Orchestrationprofessorpalmer/Puppetmaster467—~3.2kAutomated safety check: PassMIT

Similar skills

  • Claude Automation Recommender

    anthropics/claude-plugins-official

    Official

    Scans a codebase and suggests which Claude Code hooks, subagents, skills, plugins and MCP servers fit its stack, without changing any files.

    38k GitHub starsUsed in 3 repos~2.7k tokens
    Agent WorkflowsAuto-check: notes
  • CC Workflow Studio AI Editor

    breaking-brake/cc-wf-studio

    Creates and edits visual agent workflows in CC Workflow Studio through conversation, with the agent reading and writing the canvas over MCP.

    5.4k GitHub stars~561 tokensUpdated 2 days ago
    Agent WorkflowsAuto-check passed
  • Agent Deck

    asheshgoplani/agent-deck

    agent-deck, the terminal session manager for AI coding agents.

    1k GitHub stars~1.7k tokensUpdated 3 days ago
    Agent WorkflowsAuto-check passed
  • Bootstrap

    infragate/capa

    Capify an existing project. An agent skill from infragate/capa.

    724 GitHub stars~5.9k tokensUpdated 5 days ago
    Agent WorkflowsAuto-check passed
  • Puppetmaster Agent Orchestration

    professorpalmer/Puppetmaster

    Operates and supervises Puppetmaster, a multi-agent orchestrator, through its MCP tools or CLI, picking the right verb for edits, reviews, audits and long-running jobs.

    467 GitHub stars~3.2k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Agentforce Generate

    SalesforceAIResearch/agentforce-adlc

    Build, modify, audit, repair, optimize, debug, and deploy agents with Agentforce Agent Script.

    114 GitHub starsUsed in 1 repo~7.4k tokens
    Agent WorkflowsAuto-check passed

More from FerroxLabs/ijfw

All 38 skills in this repo
  • Ijfw Agents Md

    FerroxLabs/ijfw

    Maintain canonical AGENTS.md (open spec). An agent skill from FerroxLabs/ijfw.

    212 GitHub stars~2.7k tokensUpdated 4 days ago
    Auto-check passed
  • Ijfw Design

    FerroxLabs/ijfw

    A skill your agent uses when the user says: 'design', 'redesign', 'UI', 'UX', 'dashboard', 'page', 'component', 'make it look better', 'polish', 'pretty', 'professional', 'user experience'…

    212 GitHub stars~2.2k tokensUpdated 4 days ago
    Auto-check passed
  • A skill your agent uses when a milestone is shipping and you need to archive its artifacts, generate a summary, and seed the next milestone.

    212 GitHub stars~1.5k tokensUpdated 4 days ago
    Auto-check passed
  • Ijfw Critique

    FerroxLabs/ijfw

    Challenge decisions, surface counter-arguments, flag assumptions.

    212 GitHub stars~1.2k tokensUpdated 4 days ago
    Auto-check passed
  • Ijfw Cross Audit

    FerroxLabs/ijfw

    Generate a cross-platform multi-model audit (Trident) on a diff, brief, or artifact.

    212 GitHub stars~594 tokensUpdated 4 days ago
    Auto-check passed
  • Ijfw Debug

    FerroxLabs/ijfw

    Root-cause analysis with hypothesis tracking. An agent skill from FerroxLabs/ijfw.

    212 GitHub stars~578 tokensUpdated 4 days ago
    Auto-check passed

Categories

Questions about Ijfw Verify

What does Ijfw Verify do?

A skill your agent uses when about to claim completion: 'done', 'fix complete', 'tests pass', 'build succeeded', 'shipped', 'no regressions', 'ready to merge', 'ready to ship'. Ijfw Verify is an agent skill from FerroxLabs/ijfw. Use when about to claim completion: 'done', 'fix complete', 'tests pass', 'build succeeded', 'shipped', 'no regressions', 'ready to merge', 'ready to ship'.

When should I use Ijfw Verify?

Ijfw Verify fits situations like: about to claim completion: done; build succeeded.

How do I install Ijfw Verify in Claude Code?

Run `npx skills add FerroxLabs/ijfw --skill ijfw-verify -a claude-code`. Or copy the skill folder (claude/skills/ijfw-verify in FerroxLabs/ijfw) into .claude/skills/ijfw-verify in your project. Claude Code loads it when a task matches its description.

How do I install Ijfw Verify in Codex?

Run `npx skills add FerroxLabs/ijfw --skill ijfw-verify -a codex`. Or copy the skill folder (claude/skills/ijfw-verify in FerroxLabs/ijfw) into .agents/skills/ijfw-verify in your project. Codex loads it when a task matches its description.

Can I use Ijfw Verify in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add FerroxLabs/ijfw --skill ijfw-verify -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/ijfw-verify, .gemini/skills/ijfw-verify, .github/skills/ijfw-verify and .opencode/skills/ijfw-verify in your project.

What does Ijfw Verify need to run?

Going by SKILL.md and its folder, Ijfw Verify needs the command-line tools its instructions call (npm, git, curl and node).

Does Ijfw Verify access the network?

SKILL.md contains no URLs. Its commands use npm, git and curl, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Ijfw Verify safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Ijfw Verify use?

Ijfw Verify is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Ijfw Verify use?

About 2.4k tokens (SKILL.md is roughly 9.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Ijfw Verify?

Skills that share tags, products or a category with Ijfw Verify: Claude Automation Recommender (anthropics/claude-plugins-official, 38k stars), CC Workflow Studio AI Editor (breaking-brake/cc-wf-studio, 5.4k stars), Agent Deck (asheshgoplani/agent-deck, 1k stars) and Bootstrap (infragate/capa, 724 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Ijfw Verify?

FerroxLabs (a GitHub user) maintains it in FerroxLabs/ijfw, which has 212 GitHub stars. The repository holds 38 skills in this directory. The repository was last updated on October 5, 2026.

Source: FerroxLabs/ijfw on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.