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/assafkip/kipi-system/issue-startgit clone --depth 1 https://github.com/assafkip/kipi-systemWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/assafkip/kipi-system/issue-start)<a href="https://agentmods.dev/commands/assafkip/kipi-system/issue-start"><img src="https://agentmods.dev/badge/commands/assafkip/kipi-system/issue-start.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.00010 | $0.00885 |
| Opus 5 | $0.00005 | $0.00443 |
| Sonnet 5 | $0.00002 | $0.00177 |
| Haiku 4.5 | $0.00001 | $0.00089 |
Grade A, and why
issue-start 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 5d ago.
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.
What it actually says
You are starting a DSSE issue. The issue id is: $ARGUMENTS
Execute the following, in order:
-
Run
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_runner.py" load $ARGUMENTS. The output is JSON describing the loaded spec. If the command exits non-zero, stop and report the error. Do not invent an issue spec. -
Read the full spec file at the
spec_pathreturned by step 1. -
Present a plan to the advisor (main session, in the v2 rewrite topology) via the channel:
- One line summarizing the title and priority.
- The full
allowed_fileslist exactly as loaded. State explicitly: "all edits must target one of these paths." - The full
required_checkslist. State: "these must all exit 0 before/issue-verifywill record the verified receipt." - The full
required_reviewslist. - The first concrete change you plan to make, in one sentence.
-
Decide whether to wait for advisor approval or self-progress, per the issue-approve autonomy contract:
- If this is issue 1 of the PRD: wait for the advisor (main session) to approve the plan or redirect. Plan approvals route to main, NOT to the founder. Founder gates only for charter changes (PRD-approve), scope amendments mid-issue, and PRD-split manifest commit.
- If this is issue 2..N of the same PRD AND the plan's
allowed_filesandrequired_checksmatch the PRD manifest verbatim AND no scope amendment is introduced: do NOT wait. Proceed directly to step 5. - If the plan deviates from the manifest in any way (scope amendment): wait for the founder. Surface explicitly as
[FOUNDER-INPUT-NEEDED].
-
Load the execution-discipline layer: invoke the
fable-disciplineskill by name via the Skill tool (it ships in the prd-os plugin; skill-name resolution finds the installed copy — do NOT read a source-tree path, and note${CLAUDE_PLUGIN_ROOT}here is kipi-dsse's root, not prd-os's). It is prd-os's per-edit procedure — recon before edit, verify-against-a-copy with a negative self-test, single-writer chokepoints, scar-anchored why-comments. The deterministic slice (test isolation) is enforced regardless by the fable-discipline-lint PostToolUse hook in the prd-os plugin; the skill covers the judgment slice the hook cannot see. If the skill is unavailable (prd-os not installed), say so and continue — the lint hook still gates the deterministic slice. -
Run
/issue-approve. That command flips the spec status fromopentoin-progressand arms the stop-time gate. Planning before this step is intentionally gate-exempt so a loaded spec does not spam the Stop hook while the plan is being reviewed.
Do NOT ask the founder for confirmation on:
/issue-startinvocation itself (the founder already typed it or the prior issue's closeout chained to it)- Issue 1 plan-approval (route to advisor / main session)
- Commit + push between issues (the auto-approve-git PreToolUse hook handles it; if the hook denies, founder gets the prompt naturally)
- Advancing to the next issue in the PRD chain when the prior issue closed cleanly
- Codex finding triage and disposition during /issue-closeout
Rules while the issue is loaded:
- Every edit you make must target a path that matches
allowed_files. - The PreToolUse scope hook will block out-of-scope edits automatically.
- Do not mark the issue verified, reviewed, or closed yourself. Use
/issue-verify,/issue-review,/issue-closeout. - If you discover mid-issue that the scope is wrong, update the spec file's
allowed_filesfirst, explain why, then continue.
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.
- 5d ago First seen · 53 lines · 10 tokens per session scan A 781391aa5711
issue-start is a command published in the GitHub repository assafkip/kipi-system (109 stars, last pushed today), licensed MIT. It adds 10 tokens to every session and 885 once invoked, about $0.0001 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 2026-08-30.
Other commands, from other repositories
capture
Search first: /mnemonic:search {relevantkeywords} Capture after: /mnemonic:capture {namespace} "{title}".
integrate
Wire mnemonic memory operations into other Claude Code plugins.
recall
Recall memories with filtering and search.
custodian
Unified memory custodian for audits, validation, and maintenance.
query
Structured frontmatter queries using yq.
search
Full-text search across all memories.