Conciseness

98 lines covering five capabilities with zero explanation of concepts Claude already knows — it never defines what a workflow, CLI, or worktree is. Every section carries archon-specific commands, flags, or behavioral rules (e.g., "treat `descriptionTruncated: true` as an explicit signal"); the few rhetorical flourishes ("at full price") are attached to hard rules and justify them rather than pad. This matches the lean/assumes-competence anchor, not the 4-anchor's trimmable over-explanation.

Actionability

Fully executable commands with flags for the dominant case: "archon workflow run <workflow> --branch <branch-name> \"<the work, as a clear message>\" --detach", plus "archon workflow list --json", "archon workflow list <name> --full", "archon workflow runs --json", "archon workflow wait <run-id> --json", "archon workflow get <run-id> --json". Copy-paste ready with specific guidance on when each applies ("wait" as a background task, "get" for on-demand state, not a polling loop).

Workflow Clarity

The quick spine is a clear six-step sequence with an explicit pre-launch validation checkpoint ("Check the input before spending anything... If any is missing, say which, propose a corrected input, and get the user's agreement before launching. Do not silently improve it, and do not launch anyway") and a post-run verification step ("A completed run does not mean the work succeeded. Use `workflow get <run-id> --json` for the normalized `outcome`"). Validation is present, so no cap applies; the error-recovery loop (propose corrected input → user agreement) satisfies the 5-anchor.

Progressive Disclosure

A routing table up front maps all five intents to one-level-deep reference files ("running-workflows/running-workflows.md", "manage-run/manage-runs.md", "setup-and-config/setup-and-config.md", "authoring-workflows/authoring-workflows.md", "prompting-mistakes/prompting-mistakes.md"), reinforced by a Resources section annotating each file's purpose — matching the clear-overview/well-signaled anchor. No bundle files were provided to verify, and the only nit is that authoring's "node-reference.md" appears in the routing table but not in the Resources list, which is too small to drop to the 4-anchor's "minor organization gaps".

Specificity

The description enumerates every capability concretely: "run AI workflows on a repo", "manage those runs (inspect, approve, reject, cancel, resume)", "set up Archon or change its config", "author new workflows, and improve workflow prompts" — all five capabilities with specific verbs, matching the comprehensive-coverage anchor rather than the minor-gaps anchor at 4.

Completeness

Explicitly answers both questions: what ("Drive Archon through its CLI" followed by the five capabilities) and when ("Use when the user says..." with a dozen concrete trigger phrases), plus a negative boundary ("NOT for: doing the coding work yourself — Archon delegates it to isolated runs") that sharpens the when further. Exceeds the 4-anchor, which only requires a present-but-less-explicit when.

Trigger Term Quality

Extensive natural user utterances including synonyms and variations: "use archon", "run archon", "archon workflow", "fix issue #N with archon", "have archon review this PR", "what's running / check run <id>", "approve/reject/cancel/resume that run", "set up archon", "configure archon", "change my archon config", "create a workflow", "author a workflow". Every phrase is something a user would plausibly actually say, satisfying the comprehensive-synonyms anchor.

Distinctiveness Conflict Risk

"archon" is a unique product name anchoring nearly every trigger, and the explicit "NOT for" clause fences off the main overlap risk (generic coding requests). The only archon-less triggers ("create a workflow", "author a workflow") correctly route to this skill's own authoring capability, so conflict risk is minimal — matching the clear-niche anchor rather than the minor-overlap anchor at 4.