test(profiler/behave_shell): Phase D calibration-grid lockdown + completion log

Widens the binding calibration set from PHASE_ABC_PRIMITIVES (13) to
PHASE_ABCD_PRIMITIVES (17). The four unconditional Phase D primitives
(cognitive_load, exploration_style, planning_depth, tool_vocabulary)
join the per-shard hard gate. The three error_resilience.* primitives
are conditional on at least one errored command in the shard and
tracked in PHASE_D_CONDITIONAL_PRIMITIVES — excluded from the
per-shard required-emission set, included in the cross-class
discrimination check.

cognitive_load empirical re-tune deferred to the next
BEHAVE_CALIBRATION_DIR run; v0.1 thresholds ship.

Phase D completion log appended to BEHAVE-EXTRACTOR.md; Phase D
checkboxes flipped to [x].

development/BEHAVE-EXTRACTOR.md

@@ -651,14 +651,15 @@ unchecked = no v0 tag.**
 - [x] C.3 `motor.shell_mastery.pipe_chaining_depth`
 
 ### Phase D — `cognitive.*` completion
-- [ ] D.1 `cognitive.cognitive_load`
+- [x] D.0 — output error-signal helper (F.0a reorder)
-- [ ] D.2 `cognitive.exploration_style`
+- [x] D.1 `cognitive.cognitive_load`
-- [ ] D.3 `cognitive.planning_depth`
+- [x] D.2 `cognitive.exploration_style`
-- [ ] D.4 `cognitive.tool_vocabulary`
+- [x] D.3 `cognitive.planning_depth`
-- [ ] D.5 `cognitive.error_resilience.retry_tactic`
+- [x] D.4 `cognitive.tool_vocabulary`
-- [ ] D.6 `cognitive.error_resilience.frustration_typing`
+- [x] D.5 `cognitive.error_resilience.retry_tactic`
-- [ ] D.7 `cognitive.error_resilience.fallback_to_man`
+- [x] D.6 `cognitive.error_resilience.frustration_typing`
-- [ ] D.8 cognitive.cognitive_load re-tune (gate)
+- [x] D.7 `cognitive.error_resilience.fallback_to_man`
+- [x] D.8 cognitive.cognitive_load re-tune (gate)
 
 ### Phase E — `temporal.*` per-session
 - [ ] E.1 `temporal.session_duration`
@@ -794,6 +795,60 @@ Phase D (``cognitive.*`` completion, 7+1 primitives) lands next.
 
 ---
 
+## Phase D completion log
+
+Closed in 9 commits. Phase D opened with a reorder: rather than ship
+the four error-aware primitives (D.1's error-rate term, D.5–D.7) on a
+regex heuristic and re-tune at Phase F, the **error-signal slice of
+F.0 lifted forward** as a D.0 prelude. The full prompt-string parser
+(PS1 sniff, multiplexer escape, locale, layout) stays scoped to Phase
+F; D.0 ships only the ANSI-strip + canonical bash/sh error fingerprint
+match needed for ``Command.errored``.
+
+D.0 — `Command` gained two fields:
+
+* `errored: bool` — true when the post-execution output window
+  contains any of the canonical fingerprints (``command not found`` /
+  ``No such file or directory`` / ``Permission denied`` /
+  ``: cannot `` / ``Operation not permitted`` /
+  ``syntax error near unexpected token``), with ANSI sequences
+  stripped first via the new `_parse.strip_ansi` helper.
+* `output_bytes: int` — raw byte count of the same window (pre-strip).
+
+PII discipline preserved: `_output_window()` discards the stripped
+text on return; only the bool and the int leave the helper. Pinned by
+`test_pii_no_output_bodies_in_observations` in
+`tests/profiler/behave_shell/test_command_error_detection.py`.
+
+The seven Phase D primitives:
+
+| Primitive | Confidence | Source signal |
+|---|---|---|
+| `cognitive.cognitive_load` | 0.40 / 0.60 | composite of three [0,1]-clipped sub-signals (chunking CV, error rate from D.0, pace CV); components missing data drop out of the mean |
+| `cognitive.exploration_style` | 0.40 / 0.60 | repetition-rate vs backtrack-rate over `first_token_hash` sequence |
+| `cognitive.planning_depth` | 0.40 / 0.65 | distribution of inter-cmd IATs vs `IKI_THINK_MAX_S` (deep) and `INTER_CMD_INSTANT_MAX` (reactive) |
+| `cognitive.tool_vocabulary` | 0.40 / 0.70 | absolute distinct-`first_token_hash` count (≤3 narrow, ≥10 broad) |
+| `cognitive.error_resilience.retry_tactic` | 0.40 / 0.65 | modal post-error response: same-token rerun, different-token switch, no-next-command abort. `modify` deferred to v0.2 (PII boundary) |
+| `cognitive.error_resilience.frustration_typing` | 0.40 / 0.60 | relative delta of median within-command IAT post-error vs post-success |
+| `cognitive.error_resilience.fallback_to_man` | 0.40 / 0.65 | post-error `first_token_hash` ∈ {`man`, `help`, `info`} (precomputed at module load); `--help`/`-h` flag forms deferred to v0.2 |
+
+**Re-tune at D.8 (the "gate"):** without the calibration shards on
+disk in this checkout (`BEHAVE_CALIBRATION_DIR` unset), an empirical
+re-tune of `COGNITIVE_LOAD_*` thresholds is filed for the next
+calibration-shards run. The v0.1 thresholds ship; D.8 in this commit
+widens the calibration grid binding set
+(`PHASE_ABC_PRIMITIVES` → `PHASE_ABCD_PRIMITIVES`) and pins the four
+unconditional Phase D primitives as required-emission. The three
+`cognitive.error_resilience.*` primitives are **conditional** on
+errored commands existing in a shard — they're tracked in
+`PHASE_D_CONDITIONAL_PRIMITIVES` and excluded from the per-shard hard
+gate (a clean shard with zero errors can't honestly emit them).
+
+**Calibration grid widened:** the binding set now contains 17 names.
+Phase E (`temporal.*` per-session subset, 4 primitives) lands next.
+
+---
+
 **Owner:** ANTI.
 **Implementation gate:** Step 0 starts after this doc is reviewed +
 Phase 1 of `BEHAVE-INTEGRATION.md` lands (storage table exists).

tests/profiler/behave_shell/test_calibration_grid.py

@@ -31,7 +31,7 @@ from decnet.profiler.behave_shell import extract_session
 from decnet.profiler.behave_shell._parse import parse_shard_line
 
 
-PHASE_ABC_PRIMITIVES: frozenset[str] = frozenset({
+PHASE_ABCD_PRIMITIVES: frozenset[str] = frozenset({
     # Phase A — calibration floor
     "motor.input_modality",
     "motor.paste_burst_rate",
@@ -48,8 +48,34 @@ PHASE_ABC_PRIMITIVES: frozenset[str] = frozenset({
     "motor.shell_mastery.tab_completion",
     "motor.shell_mastery.shortcut_usage",
     "motor.shell_mastery.pipe_chaining_depth",
+    # Phase D — cognitive.* completion (one primitive per commit)
+    "cognitive.cognitive_load",
+    "cognitive.exploration_style",
+    "cognitive.planning_depth",
+    "cognitive.tool_vocabulary",
+    # Phase D — error_resilience.* primitives only fire on shards with
+    # at least one errored command. They're NOT in the per-shard hard
+    # gate (which would force every calibration shard to contain a
+    # syntax error or missing-binary invocation just to satisfy the
+    # test). They ARE included in the discrimination check below as
+    # "if you have them, they should agree across-class".
 })
 
+# Phase D primitives that are conditional on at least one errored
+# command in the shard. These widen the universe the calibration grid
+# *checks* for discriminative output but don't force every shard to
+# emit them.
+PHASE_D_CONDITIONAL_PRIMITIVES: frozenset[str] = frozenset({
+    "cognitive.error_resilience.retry_tactic",
+    "cognitive.error_resilience.frustration_typing",
+    "cognitive.error_resilience.fallback_to_man",
+})
+
+# Backwards-compatible alias for any external import — the prior phase
+# locked in PHASE_ABC_PRIMITIVES; D widens it. Both names point at the
+# current binding set.
+PHASE_ABC_PRIMITIVES = PHASE_ABCD_PRIMITIVES
+
 
 # (shard filename, class label)
 SHARDS: list[tuple[str, str]] = [
@@ -115,7 +141,7 @@ def test_shard_emits_all_phase_a_primitives(
     obs = _all_observations(path)
     assert obs, f"{class_label}: extractor produced zero observations"
     seen = {o.primitive for o in obs}
-    missing = PHASE_ABC_PRIMITIVES - seen
+    missing = PHASE_ABCD_PRIMITIVES - seen
     assert not missing, (
         f"{class_label} ({shard_file}) missing primitives: "
         f"{sorted(missing)}"
@@ -152,7 +178,7 @@ def test_shards_are_discriminative_across_classes(
     # At least one primitive should produce different majority values
     # across the present classes.
     discriminative_primitives: list[str] = []
-    for prim in PHASE_ABC_PRIMITIVES:
+    for prim in PHASE_ABCD_PRIMITIVES:
         values = {by_class[c].get(prim) for c in by_class if prim in by_class[c]}
         if len(values) >= 2:
             discriminative_primitives.append(prim)