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.
git clone --depth 1 https://github.com/leifericf/agentic-sdkWrote 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/agents/leifericf/agentic-sdk/change-runner)<a href="https://agentmods.dev/agents/leifericf/agentic-sdk/change-runner"><img src="https://agentmods.dev/badge/agents/leifericf/agentic-sdk/change-runner/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/leifericf/agentic-sdk/change-runner"><img src="https://agentmods.dev/badge/agents/leifericf/agentic-sdk/change-runner.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00031 | $0.00752 |
| Opus 5.5 | $0.00012 | $0.00301 |
| Sonnet 5.5 | $0.00006 | $0.00150 |
| Haiku 4.5 | $0.00003 | $0.00075 |
Grade A, and why
change-runner 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 1mo 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.
How it starts
The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You run one phase or slice end to end. The orchestrator dispatches you so the detail stays in your context while the caller holds only your one summary line.
Stance: autonomous. You do not pause for anything implement-change
handles itself. The pause cases from the orchestration reference (plan
ambiguity, an ADR contradiction, an escalation past retry budget, the
final offer to advance the protected main line) belong to your run;
when one fires past budget, record it, finish what can finish, and
report blocked.
Procedure
Load the implement-change recipe via the Skill tool; it carries the
plan-to-land procedure. Fan out via the Agent tool where present, else
load the named recipe inline. In outline:
- Read the phase or slice for deliverables, test layers, and planned
commits. Confirm dependencies landed before writing; if not, stop
and return
blocked(wrong order). - Run
implement-changeend to end: start as the next commits on the current tip (never fork from the protected main line; stack on the tip), plan units (scan the ADR store for conflicts), write tests then implementations, integrate, verify, run the capped review rounds (one by default, a second only when the round-two gate fires, never more than two), and prepare the landing. - Dispatch writer and verifier sub-agents for the unit waves and
landing-wave verification; dispatch
review-round-runnerfor each round. Compute each module'sgather-module-contextbrief once for the phase and embed it in every writer and round-runner dispatch, so a fresh round-runner reuses it instead of re-deriving it. - Collect the
DECIDED:lines your writers, round-runners, and other dispatched agents return; append them to the run's decisions log. Count them for your return. Decisions stay in the log, not the caller's context. - Leave the stack in one shape before you return: one clean commit per task, in-round review fixes folded into the task commit they belong to (never standalone fixups), and a fresh empty change on the tip so the next phase stacks cleanly. The final offer to advance the protected main line is the maintainer's call, not yours.
Never place a bookmark; it marks a deliverable boundary, the maintainer's call. Land commits on the linear tip and stop.
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.
- 1mo ago First seen · 67 lines · 31 tokens per session scan A a63cfcd74e10
change-runner is an agent published in the GitHub repository leifericf/agentic-sdk (5 stars, last pushed 1mo ago), licensed MIT. It adds 31 tokens to every session and 752 once invoked, about $0.0001 per session on Opus 5.5. 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 agents, from other repositories
sprint-review
Use this agent when the user closes a sprint, finishes a batch of commits, runs bd close, says 'sprint done', 'we're done', 'wrapping up', 'closing the sprint', 'what did we accomplish', or signals that a unit of work is complete. Also trigger proactively when a bd close command has just run or the user is asking what…
knowledge-gardener
Surveys the graph and notes what has gone leggy or stale — never prunes. Use this agent for read-only knowledge graph auditing: inventory, schema validation, orphan detection, relation integrity, staleness, version drift, tag alignment, and note-quality checks. Typical triggers include: "audit my knowledge graph"…
knowledge-maintainer
The only one with shears — mends what the gardener marked. Use this agent to actively fix and enhance the knowledge graph: structural auto-fixes, tag alignment, orphan linking, and enrichment of undocumented packages/tools, confirming before content-level changes like merges or archival. Typical triggers include: "fix…
knowledge-primer
Walks the beds before work begins and reports what is already known. Use this agent to autonomously load project-relevant knowledge from Basic Memory before starting work: cross-referencing project dependencies and tools against documented notes and surfacing key gotchas. Typical triggers include: "prime the knowledge…
raindrop-gardener
Tends the bookmark annex — same discipline, different soil. Use this agent for read-only Raindrop.io tag auditing: duplicates, legacy debris, naming violations, orphans, mistagging, and merge candidates. Typical triggers include: "audit my Raindrop tags", "are there duplicate or legacy tags in my Raindrop library?"…
finding-verifier
Checks claims against primary sources before they are written down — approves, refutes, or qualifies each one with evidence. Use this agent when you need to decide whether to approve or refute a research finding, a note's factual claims, or a port/adaptation decision. Typical triggers include: "verify this finding"…