Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/jjilli/fable-flow/iterategit clone --depth 1 https://github.com/jjilli/fable-flowWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00045 | $0.00876 |
| Opus 5 | $0.00023 | $0.00438 |
| Sonnet 5 | $0.00009 | $0.00175 |
| Haiku 4.5 | $0.00005 | $0.00088 |
Grade A, and why
iterate scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 43 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Context
- Branch: !
git branch --show-current - HEAD: !
git rev-parse --short HEAD - Working tree: !
git status --porcelain | head -20
Your job
Run the fable-flow iteration loop for bugs the user found manually:
Bug report: $ARGUMENTS
Flag: --no-review skips the reviewer pass (for trivial fixes).
This works on the current branch — typically a fable-flow/<slug> integration branch, but any branch is fine. Precondition: clean working tree (otherwise stop and tell the user what's uncommitted — never stash or discard their work yourself).
Load context first: .fable-flow/plan.md (its Contracts section still governs), .fable-flow/task.md, previous .fable-flow/iterations/*.md, and .fable-flow/memory/lessons/ — past lessons often point straight at the cause.
For each reported bug, in order:
- Reproduce before touching anything. Find or write the command/test that demonstrates the bug and run it; quote the failing output. If you cannot reproduce it, stop on that bug and report exactly what you tried and what detail would let you (that's input only the user can provide) — never shotgun-fix an unreproduced bug.
- Root-cause it. The reproduction plus the plan tells you where to look. Distinguish "the code violates the plan" from "the plan or a contract was itself wrong" — say which, because a contract-level fix must be recorded in the plan, not patched around.
- Fix it. Small fixes directly; larger ones via a
fable-flow:implementergiven a narrowed fix manifest and the current branch's HEAD as its base, then merge its branch. If a contract had to change, update the Contracts section of.fable-flow/plan.mdto match reality. - Prove it. Three things, with real output quoted: the step-1 reproduction now passes; a new or extended regression test fails on the pre-fix code and passes now; the plan's Integration verification still passes.
Bugs whose fixes are clearly independent and touch disjoint files may be fixed in parallel via implementers (same ownership rule as the main pipeline). When in doubt, sequential — an interleaved half-fix costs more than a slow one.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 43 lines · 45 tokens per session scan A 0ba258172245
iterate is a command published in the GitHub repository jjilli/fable-flow (2 stars, last pushed 1mo ago), licensed MIT. It adds 45 tokens to every session and 876 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
hire-debrief
Run the post-loop decision debrief — facilitator script for the 45-60 minute session. Surface inter-rater drift, anchor every claim to the rubric, name bias patterns out loud, render the hire-or-no-hire decision per the pre-committed rule, and capture learning for the next loop. Structured scores submitted BEFORE…
handover
End-of-session handover with wisdom capture. Writes status + reflections to docs/ops/, Starlight Vaults, and optionally GitHub.
lavra-checkpoint
Save session progress by filing beads, capturing knowledge, and syncing state.
guide
Interactive guide to fellowship. Walks you through a real task using the structured research-plan-implement flow, then shows you what's next.
tutorial
Command "tutorial" from usurahi-project/usurahi, covering /tutorial, すること, 案内する要点 and 出力方針.
explain
Generate an interactive one-module explainer for a topic — Riko gathers code scope, Senku plans a 3-5 screen teaching arc, Speedwagon authors the HTML, assembler produces explain-out/index.html.