/ijfw-verify
Run the IJFW workflow Verify phase (Deep D5). Full audit against the original brief. Usage: /ijfw-verify
$ npx -y skills add FerroxLabs/ijfw --agent claude-codeHow it fires
How this command gets triggered: by you, by Claude, or both.
- Fires itselfClaude auto-loads it when your prompt matches the work.
- You can call itInvoke it directly when you want it.
- Slash command
/ijfw-verify
Context preview
What this command does when you run it.
Run the IJFW workflow Verify phase (Deep D5). Full audit against the original brief. Usage: /ijfw-verify
Command definition
ijfw-verify.mdname: ijfw-verify
description: "Run the IJFW workflow Verify phase (Deep D5). Full audit against the original brief. Usage: /ijfw-verify"
**Ledger gate (run first):** Before emitting any VERIFY PASS, read the execute-issues ledger:
read_issues() {
local f=".ijfw/state/execute-issues.json"
[ -f "$f" ] || { printf '{"issues":[]}'; return; }
cat "$f"
}If any entry has `status: unresolved` (any `kind`: `task-incomplete`, `task-stagnated`, `unsafe-verify`, `plan-review`), emit ISSUE and halt:
ISSUE: unresolved-execute-issues
count: <N>
ids: [iss_001, iss_003]
action: resolve with /ijfw-execute resolve <id> <note> before verifying
Missing file = zero issues (day-1 fresh-install protection). Do not crash.
**Plausibility is not correctness.** Run the full Verify phase of the IJFW workflow. This is the quality gate before shipping -- a structured audit of everything built against the original brief. Every claim must trace to a command output, a test pass, or a manual verification you actually performed; never report "done" based on a plausible-looking diff alone.
**What runs:**
- Functional checks: all tests pass, startup verified, offline behavior tested
- Behavior diff against `main`: when relevant, run the same scenario on `main` and on the working tree, compare outputs side by side. Catches silent regressions that pass tests but change observable behavior. Skip for purely additive changes (new files, new endpoints) where there is no `main` baseline to diff against.
- UX audit: first impression, no "feel stupid" moments, every error has a recovery path
- Security and quality: auth on every endpoint, parameterised queries, no leaked secrets,
accessibility, linter clean
- Brief alignment: every success criterion from Discovery is checked explicitly
**Multi-AI Quality Trident (Donahoe P9):** For critical work, Verify also prepares a cross-audit document summarizing what was built and key risk areas, ready for review in Gemini, Codex, or another AI session. Use `/cross-audit` to generate it explicitly.
This command invokes `ijfw-workflow` at the D5 Verify phase directly. IJFW owns the audit loop end-to-end.
**Natural triggers:** "verify the work", "run the audit", "check everything", "quality gate", "are we ready to ship?"
If gaps are found, Verify stays open until they're resolved -- it does not auto-advance to Ship on a partial pass.
**GATE:** Verify phase ends at the SHIP GATE -- original brief re-read, what was built matches what was asked. All items pass before advancing to Ship.
**Confidence declaration (required at end of Verify):** Every Verify finding is tagged with one of:
- **VERIFIED** -- command was run, raw output is available, anyone can reproduce it.
- **LIKELY** -- reasoning provided (code read, docs consulted), not externally
verified this session.
- **GUESSING** -- insufficient information, best guess only.
- **ISSUE** -- blocker or bug surfaced; stop and document rather than advance.
Verify does not auto-advance to Ship with any finding tagged GUESSING or ISSUE. VERIFIED findings clear the gate; LIKELY findings require user acknowledgement. This is the scoped adoption of Damir Zorcic's "Declare Confidence" law, applied at ship-gate boundaries where the rigor earns its weight.
Read more
name: ijfw-verify description: "Run the IJFW workflow Verify phase (Deep D5). Full audit against the original brief. Usage: /ijfw-verify"
**Ledger gate (run first):** Before emitting any VERIFY PASS, read the execute-issues ledger:
read_issues() {
local f=".ijfw/state/execute-issues.json"
[ -f "$f" ] || { printf '{"issues":[]}'; return; }
cat "$f"
}If any entry has `status: unresolved` (any `kind`: `task-incomplete`, `task-stagnated`, `unsafe-verify`, `plan-review`), emit ISSUE and halt:
ISSUE: unresolved-execute-issues count: <N> ids: [iss_001, iss_003] action: resolve with /ijfw-execute resolve <id> <note> before verifying
Missing file = zero issues (day-1 fresh-install protection). Do not crash.
**Plausibility is not correctness.** Run the full Verify phase of the IJFW workflow. This is the quality gate before shipping -- a structured audit of everything built against the original brief. Every claim must trace to a command output, a test pass, or a manual verification you actually performed; never report "done" based on a plausible-looking diff alone.
**What runs:**
- Functional checks: all tests pass, startup verified, offline behavior tested
- Behavior diff against `main`: when relevant, run the same scenario on `main` and on the working tree, compare outputs side by side. Catches silent regressions that pass tests but change observable behavior. Skip for purely additive changes (new files, new endpoints) where there is no `main` baseline to diff against.
- UX audit: first impression, no "feel stupid" moments, every error has a recovery path
- Security and quality: auth on every endpoint, parameterised queries, no leaked secrets,
accessibility, linter clean
- Brief alignment: every success criterion from Discovery is checked explicitly
**Multi-AI Quality Trident (Donahoe P9):** For critical work, Verify also prepares a cross-audit document summarizing what was built and key risk areas, ready for review in Gemini, Codex, or another AI session. Use `/cross-audit` to generate it explicitly.
This command invokes `ijfw-workflow` at the D5 Verify phase directly. IJFW owns the audit loop end-to-end.
**Natural triggers:** "verify the work", "run the audit", "check everything", "quality gate", "are we ready to ship?"
If gaps are found, Verify stays open until they're resolved -- it does not auto-advance to Ship on a partial pass.
**GATE:** Verify phase ends at the SHIP GATE -- original brief re-read, what was built matches what was asked. All items pass before advancing to Ship.
**Confidence declaration (required at end of Verify):** Every Verify finding is tagged with one of:
- **VERIFIED** -- command was run, raw output is available, anyone can reproduce it.
- **LIKELY** -- reasoning provided (code read, docs consulted), not externally
verified this session.
- **GUESSING** -- insufficient information, best guess only.
- **ISSUE** -- blocker or bug surfaced; stop and document rather than advance.
Verify does not auto-advance to Ship with any finding tagged GUESSING or ISSUE. VERIFIED findings clear the gate; LIKELY findings require user acknowledgement. This is the scoped adoption of Damir Zorcic's "Declare Confidence" law, applied at ship-gate boundaries where the rigor earns its weight.
IJFW — It Just F*cking Works. Ferrox Labs' local-first infrastructure for AI coding agents: shared memory, smart routing, multi-AI cross-audits, disciplined workflow.
Repo: FerroxLabs/ijfw
Other commands on ijfw.
- /compress
Compress a memory/context file into terse form. Saves ~40-50% tokens per session. Usage: /compress <filepath>
Open command - /consolidate
Dream cycle -- promote patterns, prune stale, reconcile contradictions, optionally promote to global.
Open command - /cross-audit
Second-model review. Picks an auditor (codex/gemini/opencode/aider/copilot), excludes the caller, writes a prompt to paste, reads the response back. Usage: /cross-audit [--with <id> | list | compare] <target>
Open command - /cross-critique
Adversarial multi-angle critique. All three auditors fire in parallel (codex=technical, gemini=strategic, claude=ux). Counter-args ranked by rebuttal survival score, not raw severity. Usage: /cross-critique [--with <id> | list | compare] <target>
Open command - /cross-research
Two-phase multi-model research. Phase A fans codex+gemini in parallel (benchmarks + citations angles); Phase B synthesises via fresh Claude session. Usage: /cross-research [--with <id> | list | compare] <target>
Open command - /doctor
Run IJFW health check (files, MCP server, hooks, memory, caps, framing)
Open command

