Conciseness

The body is mostly operational guidance but noticeably padded: the pipeline is described three times (Lifecycle Position, When to Use the Full Pipeline, and Pipeline position), the extended surfing metaphor ("Paddling in", "The peak", "The shoulder", "The close-out") is conceptual overhead, and the "Handoffs become learning signals" section opens with "You don't need to do anything special". It sits at the mostly-efficient-but-could-be-tightened anchor rather than anchor 2 because it never explains concepts Claude already knows.

Actionability

Guidance is fully executable: exact commands (`entire status`, `ls .context-surfing/handoff-*.md`, `touch '<handoff-path>.consumed'`), a copy-paste-ready hook JSON config, a complete handoff file template with every section specified, and an explicit subagent brief for the cold-context review. For an instruction-only skill, all common cases (Entire present/absent, intent frame present/absent, drift vs. clean completion) are covered.

Workflow Clarity

Multi-step processes are clearly sequenced with explicit validation checkpoints: Activation → Drift Detection (strong/weak/not-drift taxonomy) → Recovery Protocol with a bounded re-anchor limit ("at most one local re-anchor and one cold-context check") → Exit Protocol (stop, write handoff, notify) → next-session resume steps. The pre-commit anchor check (quote verbatim → state pending action → verify trace-back) and the handoff template act as checklists with feedback loops for error recovery.

Progressive Disclosure

The bundle contains a single script (scripts/handoff-checker.sh) which is correctly referenced by the hook config, but the body itself is a ~415-line monolith: content that would fit reference files — the intent-framed-agent relationship and cadence-separation details, the handoff template, and the interoperability/pipeline listings — is inlined in SKILL.md. Structure is good (clear section headers), matching the some-structure anchor rather than anchor 4's "most content is appropriately placed".

Specificity

"Monitors context window health", "one cold-context review when self-recovery is uncertain", and "a clean handoff on real drift" list several concrete actions with only minor gaps — "bounded external anchors" is jargon rather than a user-recognizable concrete action, which keeps it below the comprehensive-coverage anchor (5).

Completeness

Both parts are explicit: the what ("Monitors context window health... Uses bounded external anchors, one cold-context review... and a clean handoff") and the when ("during large, long-running, multi-session, or explicitly context-sensitive work") plus a negative trigger. Not anchor 5 because the when-clause uses task-size categories rather than concrete user-mention trigger phrases like "when the user mentions drift or session handoffs".

Trigger Term Quality

"large", "long-running", "multi-session", "explicitly context-sensitive", and "context window" are phrases a user with this need would plausibly say, giving good keyword coverage. A few natural variations (e.g., "context is degrading", "session handoff", "long task") are missing, so it falls between anchors 3 and 4, noticeably above the midpoint.

Distinctiveness Conflict Risk

"Monitors context window health during large, long-running, multi-session, or explicitly context-sensitive work" carves a clear niche, and "Do not auto-activate merely because a medium task has an intent frame and plan" actively reduces false triggers. Minor overlap risk remains with closely related scope-drift/session-management skills that the description does not disambiguate, so it is not anchor 5.