token-pilot 0.28.3 → 0.30.0

package/agents/tp-commit-writer.md

@@ -8,8 +8,8 @@ tools:
   - mcp__token-pilot__test_summary
   - mcp__token-pilot__outline
   - Bash
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 559a0b61d20974bf33e35bc4c80dcf1b41d10d4df46cf9d05d3d5620713cd46f
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: b6831f11c61a9b255c2b6ffa04837130242fd02843463a7d30f109c1a06b3e3f
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -18,6 +18,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: commit-message authoring.

package/agents/tp-context-engineer.md

@@ -13,8 +13,8 @@ tools:
   - Edit
   - Glob
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 8977f452021085a9ba63338bf94e8903e56b30e199dc32e41acc4ec3173a931d
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 43f9364ce722ff76daf0f8720ddaf9f77e18d4c4ed8bee3e15f12d207798e778
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -23,6 +23,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: curate what AI agents see so output quality stays high.

package/agents/tp-dead-code-finder.md

@@ -11,8 +11,8 @@ tools:
   - Grep
   - Read
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 33798b70002a206c4547d08ff46caefe6dbe5a9300f94ab5dad4a57ab5fb4478
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 386760aed26df6c3595d3267954605565fad08afa8761e016079ae60c19887a8
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -21,6 +21,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: safe dead-code detection.

package/agents/tp-debugger.md

@@ -12,8 +12,8 @@ tools:
   - Read
   - Bash
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: ada78a5a3f029721fa51e7cd203395ff0e87f0ab614cc7cf0d5bcc1bf9a80435
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 71738830d025e86c70988e046a2f7f30b4590f3d284291a18609ed5fdd732321
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -22,6 +22,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: bug diagnosis via systematic triage.

package/agents/tp-dep-health.md

@@ -9,8 +9,8 @@ tools:
   - Bash
   - Read
 model: haiku
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 6224d989835ea284985b474005b8b46052b7007c4610e661b10658286b5c6624
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 12634cd28889d0a0ef1b4a6b994ba978353e14f3cb349011c393076e7e2b5c96
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -19,6 +19,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: dependency health audit.

package/agents/tp-doc-writer.md

@@ -13,8 +13,8 @@ tools:
   - Edit
   - Glob
 model: haiku
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 72347b06aaea75ed960972e96e2523c221b2ea7c892a3931aa0e7c32e4c86555
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 8e29d07dd8f58adeb9530ec477a59a6e42de6c624f322d2c6cfa8da66456b46a
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -23,6 +23,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: documentation author — decisions, ADRs, READMEs, API docs.

package/agents/tp-history-explorer.md

@@ -10,8 +10,8 @@ tools:
   - Bash
   - Read
 model: haiku
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: b2daca007e959eaf26bf9a4d92ba36c3aa277a51de4ca4db674833d36acbe11b
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 260197bc31531352f5eda3b70cf114c7c57bb7e9373f68ca76161dd68a804b0d
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -20,6 +20,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: git-history archaeology — why, when, by whom.

package/agents/tp-impact-analyzer.md

@@ -12,8 +12,8 @@ tools:
   - mcp__token-pilot__read_symbols
   - Read
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 0be2620ce0303f912f6b3334f261d169f064970c0d16602fa1e76db4cb2ea441
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 1da6936cc117a7627640fae3cc85bf13a17f0b0b0d0d533423dfb4b7c0b4b1c2
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -22,6 +22,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: impact analysis.

package/agents/tp-incident-timeline.md

@@ -8,8 +8,8 @@ tools:
   - mcp__token-pilot__read_symbol
   - Bash
 model: inherit
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 420ffc423c7479a8d4e1b226cf73eb98d6d41388317c74a950d7f3b6240b6786
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 213746bab7acb6730a6edb16e1ff7b2c56572c3adf4f94990799f1c168cfa2ad
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -18,6 +18,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: incident post-mortem timeline builder.

package/agents/tp-incremental-builder.md

@@ -13,8 +13,8 @@ tools:
   - Edit
   - Bash
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 9cb0bdf6e209d8ac613487385c01ef269d827dc3eddaf81b8eba581a3150b1e3
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 14c9adcabfb772c77a467a5fbfa682abbd5adc87e22d7fbe5d1329ffd790dde5
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -23,6 +23,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: incremental feature implementation with slice-by-slice discipline.

package/agents/tp-migration-scout.md

@@ -11,8 +11,8 @@ tools:
   - Grep
   - Glob
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: cf32cdee777430ecc6732db32b3f883a685c8a02b6dc93379d71b15555e79b3e
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 62893e448e943d0e1b928a670823ec3e152de395e487564862f145bd82161fcb
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -21,6 +21,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: migration impact mapping.

package/agents/tp-onboard.md

@@ -10,8 +10,8 @@ tools:
   - mcp__token-pilot__smart_read
   - mcp__token-pilot__smart_read_many
   - mcp__token-pilot__read_section
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: ae0b86eaffaf34bf283b94b5572481fa8c2d6a2a25193f1173b70bef0fbe1919
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 4e82f7b3c6446663e958fb6bf5eb5348bbdf33389269c888ce0dab766e50561f
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -20,6 +20,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: repository onboarding.

package/agents/tp-performance-profiler.md

@@ -11,8 +11,8 @@ tools:
   - Bash
   - Read
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 14b6fb4423a839c119120c2ea12c9dd6ab6ad1aeb13df1e7c22807b290cf1f9c
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 8b9f454a47e57e3761668de788850ef97d5d6f127b059cf8e0cef03deaca3f98
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -21,6 +21,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: performance diagnosis and targeted optimization.

package/agents/tp-pr-reviewer.md

@@ -11,8 +11,8 @@ tools:
   - mcp__token-pilot__read_for_edit
   - Read
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 73ba5844c8354088dcb10c671622daecc0e8589568de15a6001e1cf951eea586
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 91003b244472c4e65d840b55474a86ce04fba379859d588cc0fa54850b0e1e4f
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -21,6 +21,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: PR / diff review across five axes.

package/agents/tp-refactor-planner.md

@@ -8,8 +8,8 @@ tools:
   - mcp__token-pilot__outline
   - mcp__token-pilot__read_symbol
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: dcc2c2aaeb443cc9688639b4337c6069b9d5bf21e3ed757fc8b3ac8a9d61bc03
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 45f972c6b36929491a529322bac3c34fd44872f7be4a974d25c7e27cb12e9dc3
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -18,6 +18,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: refactor planning with behaviour-preservation discipline.

package/agents/tp-review-impact.md

@@ -9,8 +9,8 @@ tools:
   - mcp__token-pilot__module_info
   - Bash
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 72b635f511492188587d6cb6fd70f936ae34cf5df1f9cd9eff7849cf1231e185
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 3c1c66f952ac63a5936bec86fefda8c842fb9713bca81e48ca5bb568ccb5f367
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -19,6 +19,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: pre-merge blast-radius review.

package/agents/tp-run.md

@@ -16,8 +16,8 @@ tools:
   - Glob
   - Bash
 model: haiku
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: d665d57085db38077d0eeab74bda8bdb84c9ad59688495486059af5d3fac67cf
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: de342efe1e3ee265df1773ebde1241555750ab17de249190a5c1c200f1f8f51a
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -26,6 +26,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: general-purpose token-pilot workhorse.

package/agents/tp-session-restorer.md

@@ -9,8 +9,8 @@ tools:
   - mcp__token-pilot__session_budget
   - Bash
   - Read
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 35b7f333a28c94e7dc89fcc3171703c4b466225f55cd5c701b7592f4f6486440
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: d031f30e9cc4ea454aa256427659ed27249d820b75dc8b9b99c81ba7635230a7
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -19,6 +19,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: session-state rehydration.

package/agents/tp-ship-coordinator.md

@@ -11,8 +11,8 @@ tools:
   - Read
   - Grep
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: e8f9c28da23e318328f5afd85b09e8e7b96e0dab21a4c6779ba798cd709ced64
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 6b1c27b3dc4fad622cebff7c49e079fc764ca0ae57ef5bc4e61b563d8321092d
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -21,6 +21,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: pre-production readiness coordinator.

package/agents/tp-spec-writer.md

@@ -9,8 +9,8 @@ tools:
   - Read
   - Write
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: ed0b9f938c152c0d7be5a6a5eaf3c97c19b27ae4a9540aec342f0edb0927cb27
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 4ae44482db80a8a3a43794c6ecb665ec0b5385a274e1e5b2e3a404956075be88
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -19,6 +19,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: pre-code specification author.

package/agents/tp-test-coverage-gapper.md

@@ -10,8 +10,8 @@ tools:
   - mcp__token-pilot__test_summary
   - Glob
   - Grep
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: cc3d1f46fdb95ac3caf9344f69f1ddcd5ce5a175ee70aa150b7f9fda93edb152
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 6d862d1bcaeda3fb13099f51e40faaaf45d16d7d41d1b938609500192aa606f2
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -20,6 +20,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: test coverage gap finder.

package/agents/tp-test-triage.md

@@ -8,8 +8,8 @@ tools:
   - mcp__token-pilot__find_usages
   - mcp__token-pilot__read_symbol
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 255912c47661d203c8f9a735237bc419f97e937f788a01811bbe126ee3dd5878
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: f4e0dcbd2b4e8648efcafc9d53101a66bf394d7c90e97df7581ac47fcfbff5cb
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -18,6 +18,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: test-failure triage.

package/agents/tp-test-writer.md

@@ -13,8 +13,8 @@ tools:
   - Edit
   - Bash
 model: sonnet
-token_pilot_version: "0.28.3"
-token_pilot_body_hash: 96211a3e7f6b52dd47fef286eec3584b1c269fb3464c1102f8b7edbe470700e6
+token_pilot_version: "0.30.0"
+token_pilot_body_hash: 960fe9e907e9c7d13b14dcc22af99e8cc7e7335f99791fa808df76ac21e1f5e9
 ---
 
 You are a token-pilot agent (`tp-<name>`). Your defining contract:
@@ -23,6 +23,8 @@ For every file in a programming language, you MUST use the token-pilot MCP tools
 
 If any MCP tool fails, fall back sensibly (another MCP tool → bounded Read → pass-through) and note the fallback in your output. Never silently abandon the contract.
 
+For heavy Bash operations (test runs, builds, recursive searches, network calls, any command with potentially large stdout): when `mcp__context-mode__execute` or `ctx_batch_execute` is available, use it instead of raw Bash. Context-mode runs commands in a sandbox and only the result enters your context — typically 95% token reduction vs raw stdout dump. This is complementary to token-pilot: we own code reading, context-mode owns command execution.
+
 Your specific role is defined below.
 
 Role: targeted test authoring with TDD discipline.

package/dist/cli/tool-audit.d.ts

@@ -22,6 +22,11 @@ export interface ToolAuditRow {
     /** Calls where the recorder claimed NO savings (pass-through) — separate so
      *  they don't poison the reduction average. */
     noneCalls: number;
+    /** Calls where the MCP response was served from the session cache (the model
+     *  replayed cached tokens).  These contribute to `saved` but the mechanism
+     *  is token re-use, not structural compression — useful to split out so the
+     *  "Est.Saved*" column is understood correctly. */
+    cacheHitCalls: number;
     /** True when reduction is below the low-value threshold AND we have enough
      *  samples (≥5) to make a claim — avoids flagging tools after 1 bad run. */
     lowValue: boolean;

package/dist/cli/tool-audit.js

@@ -24,12 +24,15 @@ export function aggregateToolCalls(events, lowValueThreshold = 20, minSamples =
             tokensReturned: 0,
             tokensWouldBe: 0,
             noneCalls: 0,
+            cacheHitCalls: 0,
         };
         row.count++;
         row.tokensReturned += e.tokensReturned;
         row.tokensWouldBe += e.tokensWouldBe;
         if (e.savingsCategory === "none")
             row.noneCalls++;
+        if (e.sessionCacheHit)
+            row.cacheHitCalls++;
         byTool.set(e.tool, row);
     }
     const rows = [];
@@ -47,6 +50,7 @@ export function aggregateToolCalls(events, lowValueThreshold = 20, minSamples =
             saved,
             reductionPct,
             noneCalls: r.noneCalls,
+            cacheHitCalls: r.cacheHitCalls,
             lowValue,
         });
     }
@@ -74,7 +78,7 @@ Run a few MCP tool calls from your AI client, then re-run \`npx token-pilot tool
     lines.push(`Token Pilot — tool audit`);
     lines.push(`  ${opts.totalEvents} calls across ${rows.length} tools (cumulative across sessions)`);
     lines.push("");
-    lines.push("  Tool                     Calls      Saved   Returned  Reduction");
+    lines.push("  Tool                     Calls  Est.Saved*   Returned  Reduction");
     lines.push("  ─────────────────────────────────────────────────────────────────");
     for (const r of rows) {
         const tool = r.tool.padEnd(24);
@@ -91,6 +95,10 @@ Run a few MCP tool calls from your AI client, then re-run \`npx token-pilot tool
         lines.push("Low-value tools flagged above have <20% token reduction across ≥5 calls.");
         lines.push("Consider: check their `none` passthrough count, or whether a cheaper alternative (Grep, Read) would do the job.");
     }
+    lines.push("");
+    lines.push("* Est.Saved is estimated against a full-file read baseline. Actual prompt");
+    lines.push("  savings depend on client caching — use `cacheHitCalls` in --json output");
+    lines.push("  to distinguish structural compression from cache re-use.");
     return lines.join("\n");
 }
 export async function runToolAudit(opts) {

package/dist/core/policy-engine.d.ts

@@ -6,8 +6,6 @@
 export interface PolicyConfig {
     /** Advisory hints when an expensive tool is used where a cheaper alternative exists */
     preferCheapReads: boolean;
-    /** Track if read_for_edit was called before edit (advisory) */
-    requireReadForEditBeforeEdit: boolean;
     /** Always cache project overview in session cache */
     cacheProjectOverview: boolean;
     /** Warn after N full-file reads in a session */
@@ -25,13 +23,11 @@ export declare const DEFAULT_POLICIES: PolicyConfig;
 export interface PolicyCheckContext {
     fullFileReadsCount: number;
     tokensReturned: number;
-    readForEditCalled?: Set<string>;
-    editTargetPath?: string;
     totalCallCount?: number;
     totalTokensReturned?: number;
 }
 export interface PolicyAdvisory {
-    level: 'info' | 'warn';
+    level: "info" | "warn";
     message: string;
 }
 /**