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 skills add doodledood/manifest-dev --skill escalategit clone --depth 1 https://github.com/doodledood/manifest-devWrote 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/doodledood/manifest-dev/escalate)<a href="https://agentmods.dev/skills/doodledood/manifest-dev/escalate"><img src="https://agentmods.dev/badge/skills/doodledood/manifest-dev/escalate/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/skills/doodledood/manifest-dev/escalate"><img src="https://agentmods.dev/badge/skills/doodledood/manifest-dev/escalate.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00035 | $0.00807 |
| Opus 5.5 | $0.00014 | $0.00323 |
| Sonnet 5.5 | $0.00007 | $0.00161 |
| Haiku 4.5 | $0.00003 | $0.00081 |
Grade A, and why
escalate 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 2d 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
This is a terminal handoff, called only after useful independent work is exhausted. An earlier blocker notice belongs in the executor's progress report; it does not end the run. If useful independent work remains, return to the calling executor to continue it before escalating.
Receive the Manifest path from /do, along with the affected gates and the actual evidence behind each. Relay that evidence as supplied, without inventing verification the run did not perform. Surface it with the blocker evidence: the criterion (INV-G or AC ID) that can't be met, the credible recovery paths investigated and why they failed, and the intervention needed — unavailable knowledge or access, new authority, or a change to a binding requirement. Name what independent work completed and why the remainder depends on that intervention. Recommend a resolution; report it without starting a question-and-wait loop.
Also name material autonomous decisions and their rationale, including Appetite revisions and the benefit that justified their added complexity or maintenance; any Process Guidance departed from; any deviation from the Initial Approach or the Deliverable order; findings a gate reported below its threshold, pre-existing issues left as unrelated to the change, and the debt /do hands over, item by item as supplied, offering ticket-up for any the user wants worked; each offer flagged for a person rather than accepted — whether the run flagged it or the Manifest's ASM-* entries carry it — and each decision record written with an agent as its decider, with what accepting the offer or reviewing the record would change; and any costly-threshold concern the run recorded while continuing repairs. Include these even when the execution log already holds them. Carry any drafted Taste entry /do supplies verbatim, with its question for the user's yes.
The user's call. A decision is the user's call when two careful people with
the same evidence could reasonably decide it differently, and the difference
would matter to the user. Test each material decision against that sentence;
your own confidence does not settle it. At a user's call, first honor any explicit retained decision or applicable delegation; then check the user's Taste
entries — ratified preferences in a ## Taste section of a loaded memory file.
Follow an entry only where it clearly bears: it speaks to this kind of call, and
following it would change the choice. A general lean that would tilt almost any
call, such as less effort or lower cost, does not bear on its own. Otherwise
decide on the evidence. Wherever decisions are reported, list the user's calls
first, each with the alternative not taken and what decided it — the entry
followed, or that no entry applied.
A false binding premise outside the caller's amendment authority can arrive here even when its gate passed. Quote the report or name what execution showed in place of attempts a passing gate does not have. A costly bar alone is not a blocker while repairs remain viable; it is reported as a concern, never lowered by this skill.
A BLOCKED verifier verdict is not sufficient on its own: the executor first investigates what it can recover and completes useful independent work. Quote the remaining BLOCKED note and name the required outside action. A pure external wait reports as pending under a no-wait policy; otherwise the executor waits and re-verifies. A stalled wait with no viable recovery path names the intervention needed here. Pure user questions about the manifest or process are answered inline by the executor.
If the user later supplies a scope or requirement change, invoke define
with the Manifest path, then resume the executor. If they clear the blocker
or supply the missing context, resume it directly. Their response is steering;
resumed execution remains unattended.
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.
- 2d ago Changed f4ce10661e5a
- 12d ago Changed 07218d1fb353
- 13d ago Changed · +12 lines 4bfb7a385e83
- 1mo ago Changed 1a3c091dc9ce
- 1mo ago Changed · -1 lines · -3 tokens per session 37b1d7c57272
- 1mo ago Changed · +19 lines · -23 tokens per session 0e5fd6edc917
- 1mo ago Changed · -2 lines 4944c5ebd5ee
- 1mo ago First seen · 18 lines · 61 tokens per session scan A 085f6084f44f
escalate is a skill published in the GitHub repository doodledood/manifest-dev (76 stars, last pushed yesterday), licensed MIT. It adds 35 tokens to every session and 807 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 2026-08-30.
Other skills, from other repositories
vertical-onboarding
Onboarding-and-switching playbook for SMB Product-Builder products. Defines the first-run experience that turns a prospect leaving an incumbent (ServiceTitan/Toast/Mindbody/Shopify/QuickBooks) into an activated user — import-first onboarding, the activation milestone, sample-data fallback, and the time-to-first-value…
map
Ask which orc command, skill, or flow fits your situation — the router over everything orc ships.
wait-what
Stop and re-pitch — the last message did not land; say it again simply, in the project's own vocabulary.
status
Use this skill when the user asks "where am I", "what's the current task", "what's blocking me", "show status", "what needs sign-off", or "what's next". Prints a snapshot of the active plan, gate, sign-off progress, and next action. Reads only — writes nothing.
handoff
Compact the current conversation into a redacted handoff document, saved to the OS temp dir, that a fresh agent in another session or tool can pick up — /orc:resume is the counterpart for resuming in-repo orc state.
collaboration
Use for /collaboration, collaboration repair, friction signals, soft-skills coaching, or structured human-Claude interaction techniques such as compromise, rubber ducking, energy checks, pairing, retros, working agreements, and perspective swaps.