chore(release): bump to v0.25.6

Author: claude

.agents/skills/pk-explain-routing/SKILL.md

@@ -0,0 +1,6 @@
+---
+name: pk-explain-routing
+description: "Use the model-recommender skill to explain the model-routing trace for $ARGUMENTS."
+---
+
+Use the model-recommender skill to explain the model-routing trace for $ARGUMENTS.

.agents/skills/team-create/SKILL.md .agents/skills/pk-team-create/SKILL.md.agents/skills/team-create/SKILL.md renamed to .agents/skills/pk-team-create/SKILL.md

@@ -1,9 +1,9 @@
 ---
-name: team-create
-description: "Command: team-create"
+name: pk-team-create
+description: "Command: pk-team-create"
 ---
 
-# Command: team-create
+# Command: pk-team-create
 
 Full team derivation from scratch. Writes 8 Role + 8 Actor + 8 Binding
 entities, rewrites `context/team/roster.md`, and emits a chartering
@@ -12,7 +12,7 @@ DecisionRecord that supersedes the prior team DecisionRecord.
 ## Syntax
 
 ```
-team-create
+pk-team-create
   --subscription <provider>:<tier>   # e.g. anthropic:max-5x
   --providers <list|"any">           # comma-separated; "any" = all accessible
   --parallelism-cap <int>            # max clones per role, default 5
@@ -46,7 +46,7 @@ Three-level precedence (see `references/landscape-resolution.md`):
 If no artifact is found at any level, **abort** with:
 > "No landscape-summary artifact found. Run the Haiku landscape-ingest
 > task to produce an ART-*-LandscapeSnapshot artifact before running
-> team-create."
+> pk-team-create."
 
 If the resolved artifact is older than 90 days, emit a warning but
 continue. Record the artifact ID and its `created` date in the
@@ -162,7 +162,7 @@ If `context/team/roster.md` already exists:
 1. Parse the prior team's Actor and Binding IDs from roster.md.
 2. For each prior Binding:
    `binding-management.end_binding(id, reason=
-     "superseded by team-create run <ISO-timestamp>")`
+     "superseded by pk-team-create run <ISO-timestamp>")`
 3. For each prior Actor whose model ID does NOT appear in the new
    team assignments:
    `actor-profile.deactivate_actor(id)`
@@ -250,12 +250,12 @@ decision-record-write(
 ```
 
 The `inputs_snapshot.weights` block is the canonical weight store
-that `team-rebalance` will read on future runs.
+that `pk-team-rebalance` will read on future runs.
 
 ## Dry-run output format
 
 ```
-=== team-create DRY RUN ===
+=== pk-team-create DRY RUN ===
 Subscription: <value>   Governance floor: <value>
 Landscape: <artifact-id> (<date>)
 Weights: C=0.60 K=0.20 L=0.10 G=0.10

.agents/skills/team-rebalance/SKILL.md .agents/skills/pk-team-rebalance/SKILL.md.agents/skills/team-rebalance/SKILL.md renamed to .agents/skills/pk-team-rebalance/SKILL.md

@@ -1,18 +1,18 @@
 ---
-name: team-rebalance
-description: "Command: team-rebalance"
+name: pk-team-rebalance
+description: "Command: pk-team-rebalance"
 ---
 
-# Command: team-rebalance
+# Command: pk-team-rebalance
 
-Apply a `team-review` recommendation. Targeted re-tiering of one or
+Apply a `pk-team-review` recommendation. Targeted re-tiering of one or
 more roles without full team recreation. Requires `--confirm` to
 prevent accidental writes.
 
 ## Syntax
 
 ```
-team-rebalance
+pk-team-rebalance
   --roles <list|"all">         # comma-separated role names, or "all"
   --confirm                    # required; prevents accidental writes
   --reason <string>            # recorded in the DecisionRecord amendment
@@ -48,23 +48,23 @@ silently apply the new weights:
 ```
 WARNING: A newer DEC-*-TeamWeights record (created <ISO-date>)
 post-dates the governing team DecisionRecord (created <ISO-date>).
-Weight policy has changed since the last team-create run. This
+Weight policy has changed since the last pk-team-create run. This
 targeted rebalance uses stored weights from the governing DEC.
 To apply the updated weights across all roles, run:
-  team-rebalance --roles all --confirm --reason "<reason>"
+  pk-team-rebalance --roles all --confirm --reason "<reason>"
 ```
 
 CLI `--weight-overrides` and `--threshold-overrides` always override
 stored values regardless of DEC age.
 
-If `--roles all` is specified, re-run the full `team-create` logic
-(steps 1–8 of `team-create`) and write a new DecisionRecord instead
+If `--roles all` is specified, re-run the full `pk-team-create` logic
+(steps 1–8 of `pk-team-create`) and write a new DecisionRecord instead
 of amending. The remainder of this document covers the non-`all`
 path.
 
 ### Step 2 — Resolve landscape snapshot
 
-Same resolution logic as `team-create` Step 1. If `--landscape` is
+Same resolution logic as `pk-team-create` Step 1. If `--landscape` is
 not supplied, use the latest `landscape-summary` artifact. Warn if
 older than 90 days; do not block.
 
@@ -91,7 +91,7 @@ For each role where a model change is needed:
 ```
 binding-management.end_binding(
   id=<current-BIND-id>,
-  reason="superseded by team-rebalance: <--reason> (<ISO-timestamp>)"
+  reason="superseded by pk-team-rebalance: <--reason> (<ISO-timestamp>)"
 )
 ```
 
@@ -155,11 +155,11 @@ A new DecisionRecord is NOT written unless `--roles all` is used.
 
 ## The `--roles all` shortcut
 
-`team-rebalance --roles all --confirm --reason "<reason>"` is
-equivalent to running `team-create` with the stored weights and
+`pk-team-rebalance --roles all --confirm --reason "<reason>"` is
+equivalent to running `pk-team-create` with the stored weights and
 subscription, except:
 - A new DecisionRecord IS written (superseding the current one).
-- The deactivation sequence from `team-create` Step 5 applies in full.
+- The deactivation sequence from `pk-team-create` Step 5 applies in full.
 
 Use this when the landscape has shifted so broadly that targeted
 rebalancing would touch every role anyway.

.agents/skills/team-review/SKILL.md .agents/skills/pk-team-review/SKILL.md.agents/skills/team-review/SKILL.md renamed to .agents/skills/pk-team-review/SKILL.md

@@ -1,9 +1,9 @@
 ---
-name: team-review
-description: "Command: team-review"
+name: pk-team-review
+description: "Command: pk-team-review"
 ---
 
-# Command: team-review
+# Command: pk-team-review
 
 Read-only health-check. Compares current team assignments against the
 latest landscape snapshot and surfaces tier-drift, unavailable models,
@@ -12,7 +12,7 @@ and new outperformers. No entities are written.
 ## Syntax
 
 ```
-team-review
+pk-team-review
   [--landscape-artifact <ART-id>]  # explicit landscape override; skips discovery
   [--landscape <artifact-id>]      # alias for --landscape-artifact
   [--threshold <float>]            # flag if tier-score drifted > N pts, default 0.15
@@ -23,7 +23,7 @@ team-review
 ### Step 1 — Load current team
 
 1. Read `context/team/roster.md` to get the active role → model mapping.
-2. Resolve the governing DecisionRecord (the most recent `team-create`
+2. Resolve the governing DecisionRecord (the most recent `pk-team-create`
    DecisionRecord) via `decision-record.get_decision`.
 3. Extract stored formula weights and tier scores from
    `spec.inputs_snapshot`. These are the baseline scores for the diff.
@@ -32,7 +32,7 @@ team-review
 ### Step 2 — Load landscape snapshot
 
 Landscape resolution follows the same three-level precedence as
-`team-create` (see `references/landscape-resolution.md`):
+`pk-team-create` (see `references/landscape-resolution.md`):
 
 1. If `--landscape-artifact <ART-id>` (or alias `--landscape`) is
    supplied, load that artifact directly.
@@ -68,7 +68,7 @@ Score all accessible candidates with the stored weights. Identify any
 model that:
 - Would out-score the current assignment for its role by more than
   `--threshold` (default 0.15).
-- Appeared in the landscape since the last `team-create` run.
+- Appeared in the landscape since the last `pk-team-create` run.
 
 ### Step 5 — Detect unavailable models
 
@@ -84,8 +84,8 @@ or `major_outage`.
 Output format (stdout only — no files written):
 
 ```
-=== team-review — <date> ===
-Baseline: DecisionRecord <DEC-id> (<team-create date>)
+=== pk-team-review — <date> ===
+Baseline: DecisionRecord <DEC-id> (<pk-team-create date>)
 Landscape: <artifact-id> (<artifact date>) [STALE: >90 days if applicable]
 Weights used: C=<C> K=<K> L=<K> G=<G>
 Threshold: <threshold>
@@ -110,7 +110,7 @@ STABLE (no action needed):
 SUMMARY:
   2 roles recommended for rebalance
   1 role urgently needs replacement (major_outage)
-  Run: team-rebalance --roles developer,junior-developer --confirm \
+  Run: pk-team-rebalance --roles developer,junior-developer --confirm \
        --reason "<reason>"
 =============================
 ```
@@ -121,9 +121,9 @@ Omit empty sections. Always emit the SUMMARY even if no issues found.
 
 None. This command is fully read-only.
 
-## Distinction from team-create
+## Distinction from pk-team-create
 
-`team-review` reads and diffs; `team-create` writes. Running
-`team-review` is always safe and reversible. Only run `team-create`
-or `team-rebalance` when you have reviewed the diff and confirmed
+`pk-team-review` reads and diffs; `pk-team-create` writes. Running
+`pk-team-review` is always safe and reversible. Only run `pk-team-create`
+or `pk-team-rebalance` when you have reviewed the diff and confirmed
 the recommended changes.

.claude/commands/pk-dec-find.md

@@ -1,5 +1,5 @@
 ---
-argument-hint: ""
+argument-hint: "filter"
 allowed-tools: []
 ---
 

.claude/commands/pk-dec.md

@@ -1,5 +1,5 @@
 ---
-argument-hint: ""
+argument-hint: "decision-title"
 allowed-tools: []
 ---
 

.claude/commands/pk-note-promote.md

@@ -1,5 +1,5 @@
 ---
-argument-hint: ""
+argument-hint: "note-id"
 allowed-tools: []
 ---
 

.claude/commands/pk-observe.md

@@ -1,5 +1,5 @@
 ---
-argument-hint: ""
+argument-hint: "owner-name observation"
 allowed-tools: []
 ---
 

.claude/commands/pk-owner-bootstrap.md

@@ -1,5 +1,5 @@
 ---
-argument-hint: ""
+argument-hint: "owner-name"
 allowed-tools: []
 ---
 

.claude/commands/pk-release-audit.md

@@ -1,5 +1,5 @@
 ---
-argument-hint: "[--repo-root=<path>]"
+argument-hint: "[--repo-root=<path>] [--tree=context|src-context|both]"
 allowed-tools: []
 ---