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 skills/spacedock-dev/spacedock/feedback-rejection-flownpx skills add spacedock-dev/spacedock --skill feedback-rejection-flowgit clone --depth 1 https://github.com/spacedock-dev/spacedockWrote 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/skills/spacedock-dev/spacedock/feedback-rejection-flow)<a href="https://agentmods.dev/skills/spacedock-dev/spacedock/feedback-rejection-flow"><img src="https://agentmods.dev/badge/skills/spacedock-dev/spacedock/feedback-rejection-flow.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 | $0.00067 | $0.01116 |
| Opus 5 | $0.00034 | $0.00558 |
| Sonnet 5 | $0.00013 | $0.00223 |
| Haiku 4.5 | $0.00007 | $0.00112 |
Grade A, and why
feedback-rejection-flow 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 4d 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 — 35 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Feedback Rejection Flow
This skill loads at the rejection-handling point; rejection detection, correction-record write scope, reuse conditions, and the budget probe stay always-on in the FO contract.
Feedback Rejection Flow
When a feedback stage recommends REJECTED, run these five steps in order. Each is one action with one completion condition, and is unfinished until that condition holds: never treat a step's first command as its completion.
-
Deliver the authorized correction. Read the rejected stage's
feedback-totarget — the stage that receives the fix request, not the reviewer — and the already-authorized workflow package: rejected snapshot, finding evidence, existing workflow classifications, FO-authorized dispositions, and concrete revise assignment. If the distinct authorization or assignment is missing, hold at the active workflow's review-finding checkpoint; routing is ineligible. Route the package unchanged to the target stage in the same worktree, carrying the concrete assignment, not an acknowledgment request or a new classification request. Reuse the existing handle through«addressable-worker»only when it is addressable, reuse conditions pass, and«context-budget»()reports it under budget; otherwise shut down and fresh-dispatch. If no probe is declared, proceed to reuse. Done when the correction is complete in durable workflow state: the target worker's own entries closing this round's review log where the workflow keeps one, otherwise its«completion-signal»attributed by mailbox content, task path, or durable state. The immediate routing response is never completion. -
Record this round, once. Append the authorized
### Feedback Cyclesline for this round when the active workflow declares that projection, then invoke the neutral recorder exactly once for the whole rejection cycle:${SPACEDOCK_BIN:-spacedock} gate record ENTITY --round STAGE/CYCLE --briefing PATH/briefing.json --log PATH/briefing.review.jsonl. ENTITY is the entity operand the recorder requires; omitting it is a usage error, not a recording. CYCLE is this rejection cycle's number and the only round id this cycle publishes; there is no second publication after the reviewer re-run. Do not define, normalize, or interpret the Cycle line's category labels, fields, tolerance, estimate, or drift grammar. The recorder retains the canonical two-file room and advancesreview-round, without receiving or interpreting the Cycle line. Done when the recorder exits successfully reporting the complete round summary, counting every entry this round accumulated. A usage error — the invocation itself was malformed — is not a round failure: correct the invocation and run it again, which completes this one publication rather than adding a second. If the recorder instead refuses the round, produces no result, or reports an incomplete round, hold the flow: do not claim that the round was recorded, re-run the reviewer, or prepare the next gate.
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.
- 4d ago First seen · 35 lines · 67 tokens per session scan A 9896b43ff641
feedback-rejection-flow is a skill published in the GitHub repository spacedock-dev/spacedock (76 stars, last pushed 5d ago), licensed Apache-2.0. It adds 67 tokens to every session and 1,116 once invoked, about $0.0003 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 skills, from other repositories
fleet
Parallel campaign orchestrator. Runs multiple campaigns in coordinated waves within a single session. Spawns 2-3 agents per wave in isolated worktrees, collects discoveries, shares context between waves. Use when work decomposes into 3+ independent streams that can run simultaneously.
archon
Autonomous multi-session campaign agent. Decomposes large work into phases, delegates to sub-agents, reviews output, and maintains campaign state across context windows. Use for work that spans multiple sessions and needs persistent state, quality judgment, and strategic decomposition.
setup
First-run experience for the harness. Three modes: Recommended (guided, 3 min), Full Tour (guided + skill walkthrough, 8 min), and Express (zero questions, 30 sec). Installs hooks first, detects stack, configures harness.json, runs a live demo on real code, and prints a reference card.
daemon
Continuous autonomous operation mode. Keeps campaigns running 24/7 by chaining Claude Code sessions via RemoteTrigger. Each session picks up from the campaign's continuation state, works until context runs low or the phase completes, then schedules the next session. Auto-stops on campaign completion or budget…
dashboard
Real-time harness observability dashboard. Reads campaigns, fleet sessions, telemetry, and pending queues to present a snapshot of harness state at a glance. Invoked by /dashboard, /do status, or phrases like "what's happening" and "show activity".
evolve
Research-driven multi-cycle improvement director. Forms causal hypotheses about why scores are low, validates them with scout agents before attacking, dispatches axis-parallel fleet attacks, extracts transferable patterns, and runs indefinitely within a budget envelope. Accumulates a persistent belief model and…