4207 Commits

This Branch

This Branch

All Branches

• fix(coding-agent): reuse session thinking for compaction closes #3438
  Mario Zechner · 2026-04-20 15:09:50 +02:00
  da6a81d398
• fix(coding-agent): add missing typebox dependency

  closes #3434

  Mario Zechner · 2026-04-20 15:08:34 +02:00
  d66ef6dce8
• fix(coding-agent): preserve share page tool output indentation closes #3440
  Mario Zechner · 2026-04-20 14:46:37 +02:00
  5ea8a276eb
• feat(coding-agent): make scoped models and tree filter shortcuts configurable (#3343)

  - Add app.models.{save,enableAll,clearAll,toggleProvider,reorderUp,reorderDown}
  - Add app.tree.filter.{default,noTools,userOnly,labeledOnly,all,cycleForward,cycleBackward}
  - Migrate scoped-models-selector cancel to tui.select.cancel
  - Prefer reserved actions on key collision in extension shortcut conflict check

  Marek Pazik · 2026-04-20 14:35:30 +02:00
  d4e2e563ae
• fix(coding-agent): split /clone from /fork UX

  closes #2962

  Mario Zechner · 2026-04-20 14:33:32 +02:00
  d554409b1f
• fix(coding-agent): add OpenRouter attribution headers

  closes #3414

  Mario Zechner · 2026-04-20 14:14:36 +02:00
  62c1c4031c
• feat(coding-agent): add fork position and duplicate session option (#3431)
  Armin Ronacher · 2026-04-20 14:13:34 +02:00
  23569e304b
• feat(ai): allow overriding oauth callback bind host closes #3396 (#3409)
  Michael · 2026-04-20 13:43:08 +02:00
  ebc60aa825
• fix(coding-agent): avoid duplicate symlinked skills in pi config (#3417)

  - dedupe resolved skill entries by canonical path (realpathSync) with raw-path fallback matching loadSkills()
  
  - preserve precedence by applying skill dedupe after resource precedence sorting (first winner retained)
  
  - add tests to verify expected behavior
  
  - closes #3405

  Ramiz Wachtler · 2026-04-20 13:23:00 +02:00
  2b45243817
• fix(ai): replace stale OpenRouter Llama 4 Maverick test model with Llama 4 Scout (#3421)

  * fix(ai): replace stale OpenRouter Llama 4 Maverick test model
  
  - Switch OpenRouter Meta test model from `meta-llama/llama-4-maverick` to `meta-llama/llama-4-scout` in AI tests.
  
  - CI failure body:
    - packages/ai/test/context-overflow.test.ts(531,41): error TS2345: Argument of type "meta-llama/llama-4-maverick" is not assignable to parameter of type "...".
    - packages/ai/test/total-tokens.test.ts(520,40): error TS2345: Argument of type "meta-llama/llama-4-maverick" is not assignable to parameter of type "...".
  
  * chore(ai): add changelog entry for OpenRouter test model swap
  
  - Add Unreleased Fixed entry for switching OpenRouter Meta test model to llama-4-scout
  
  - Reference PR #3421 attribution

  Ramiz Wachtler · 2026-04-20 13:21:29 +02:00
  cbe0e8304f
• chore: approve contributor cmraible
  github-actions[bot] · 2026-04-19 20:40:54 +00:00
  877279f518
• fix(ai): strip JSON Schema meta keys for Cloud Code Assist (#3412)
  Vladyslav Tkachenko · 2026-04-19 16:54:49 +02:00
  f732f5e858
• fix(coding-agent): use browser-safe share viewer shortcuts (#3374)

  fixes #3358

  Andrea Baccega · 2026-04-19 08:49:12 +02:00
  9b28e185db
• feat(amazon-bedrock): conditionally omit maxTokens from inferenceConfig (#3400)

  When model.maxTokens is 0 (unset/unknown), omit maxTokens from the
  Bedrock ConverseStream inferenceConfig instead of sending 0.
  
  Bedrock's inferenceConfig.maxTokens is optional — when omitted, the
  model uses its own internal default. This is optimal for Bedrock's
  token quota management:
  
  - At request start, Bedrock reserves input_tokens + max_tokens from
    your TPM quota
  - For Claude 3.7+ models, output tokens have a 5x burndown rate
  - Sending an unnecessarily high maxTokens wastes TPM capacity during
    the reservation window, reducing concurrent request throughput
  - Omitting it lets Bedrock use the model default (~4096 for Claude),
    matching expected output sizes and maximizing quota utilization
  
  Also applies the same conditional pattern to temperature — only
  include it in inferenceConfig when explicitly set.
  
  Changes:
  - simple-options.ts: Use ?? instead of || so explicit 0 is preserved;
    return undefined when model.maxTokens is 0 (unset)
  - amazon-bedrock.ts: Spread maxTokens and temperature into
    inferenceConfig only when defined
  
  Fixes #3399

  wirjo · 2026-04-19 08:48:43 +02:00
  a5fac1ef09
• fix(amazon-bedrock): pass model.baseUrl as endpoint to BedrockRuntimeClient (#3402)

  BedrockRuntimeClient was constructed without the endpoint option,
  causing model.baseUrl to be silently ignored. This meant custom
  Bedrock endpoints (VPC endpoints, proxy setups, custom routing)
  were never used — requests always went to the default regional
  endpoint.
  
  Pass model.baseUrl as the endpoint config when set. When not set
  (the default), the SDK falls back to the standard regional endpoint
  constructed from the region config, preserving existing behavior.
  
  Fixes openclaw/openclaw#47899

  wirjo · 2026-04-19 08:47:10 +02:00
  5a889ef58b
• chore: approve contributor Michaelliv
  github-actions[bot] · 2026-04-19 06:41:37 +00:00
  c6a493c50b
• fix(coding-agent): coerce stringified JSON edits in edit tool (#3370)

  Some models (Opus 4.6, GLM-5.1) send the edits parameter as a JSON
  string instead of a parsed array. This fails AJV validation with
  'must be array' and models fall back to sed/python.
  
  Parse stringified edits in prepareEditArguments before validation.

  Danila Poyarkov · 2026-04-18 18:29:23 +02:00
  a2ec01e12f
• fix manifest glob expansion and add regression test (#3350)
  Neon · 2026-04-18 13:16:39 +02:00
  182d4ceea3
• Add [Unreleased] section for next cycle
  Mario Zechner · 2026-04-18 01:32:50 +02:00
  00887125c0
• Release v0.67.68
  Mario Zechner · 2026-04-18 01:31:38 +02:00
  v0.67.68 a1edb8a463
• fix(ai): use SDK token auth and omit Bedrock display in GovCloud closes #3359
  Mario Zechner · 2026-04-18 01:24:13 +02:00
  454b9619c7
• fix(ai): sanitize Mistral tool schemas closes #3361
  Mario Zechner · 2026-04-18 00:54:46 +02:00
  2dddc5ba25
• feat(agent,coding-agent): per-tool executionMode override for sequential tool execution (#3345)

  * feat(agent,coding-agent): add per-tool executionMode field to AgentTool and ToolDefinition
  
  Add optional executionMode?: ToolExecutionMode to AgentTool and
  ToolDefinition interfaces. Propagate through wrapToolDefinition and
  createToolDefinitionFromAgentTool. No behavioral change yet — agent
  loop will read this field in a follow-up.
  
  * feat(agent): support per-tool executionMode override for sequential execution
  
  When a tool defines executionMode='sequential', the agent loop
  forces sequential execution of all tool calls in that batch,
  even if the global config is parallel.
  
  * feat(coding-agent): re-export ToolExecutionMode from @mariozechner/pi-agent-core
  
  Makes the type available to extensions that want to set
  executionMode on tool definitions.
  
  * feat(coding-agent): add tic-tac-toe extension example with executionMode: sequential
  
  Demonstrates per-tool executionMode: the agent plays via move/play
  tool calls that share a cursor. Without sequential execution, play
  can resolve before earlier moves finish, landing on the wrong cell.

  Aliou Diallo · 2026-04-18 00:45:57 +02:00
  bfa11a50e4
• Add [Unreleased] section for next cycle
  Mario Zechner · 2026-04-17 23:06:48 +02:00
  cbf0d1fbfa
• Release v0.67.67
  Mario Zechner · 2026-04-17 23:06:48 +02:00
  v0.67.67 1c0b628882
• fix(release): use npx shx for explicit versions
  Mario Zechner · 2026-04-17 23:06:48 +02:00
  27620f9383
• chore: allow explicit release versions
  Mario Zechner · 2026-04-17 23:06:48 +02:00
  1573d46ac2
• chore: approve contributor vekexasia
  github-actions[bot] · 2026-04-17 21:05:58 +00:00
  6121a35827
• chore(changelog): update unreleased entries
  Mario Zechner · 2026-04-17 22:54:02 +02:00
  b0c2579fd7
• fix(ai): use reasoning_effort for Mistral Small 4 closes #3338
  Mario Zechner · 2026-04-17 22:44:27 +02:00
  62778a82d5
• fix(coding-agent): prevent text selection from triggering click toggles in exported HTML (#3332)
  xu0o0 · 2026-04-17 19:23:47 +02:00
  e6f473f432
• fix(ai): preserve qwen chat-template thinking closes #3325
  Mario Zechner · 2026-04-17 16:59:00 +02:00
  e3f6912d49
• fix(coding-agent): keep scoped-model reorder inert in implicit all-enabled state closes #3331
  Mario Zechner · 2026-04-17 16:58:49 +02:00
  0bb4ccf11f
• fix(coding-agent): resolve captured commands on close closes #3027
  Mario Zechner · 2026-04-17 16:45:05 +02:00
  3cea63cf29
• fix(coding-agent): retry connection lost errors

  closes #3317

  Mario Zechner · 2026-04-17 16:45:05 +02:00
  c15e4d4913
• chore: approve contributor mpazik
  github-actions[bot] · 2026-04-17 14:06:22 +00:00
  346c82425e
• fix(coding-agent): disambiguate compact extension labels

  closes #3308

  Armin Ronacher · 2026-04-17 09:46:47 +02:00
  32a305cb9b
• chore(coding-agent): update changelog for git install devDependency omission closes #3009
  Mario Zechner · 2026-04-17 01:53:20 +02:00
  ddbf642179
• fix(ai): trust requested Codex service tier (#3307)
  Markus Ylisiurunen · 2026-04-17 01:50:04 +02:00
  2cdac7382a
• fix(coding-agent): restore afterToolCall error overrides closes #3051
  Mario Zechner · 2026-04-17 01:49:34 +02:00
  e9808b5854
• chore: remove dyyz1993 from approved contributors
  Mario Zechner · 2026-04-17 01:25:23 +02:00
  32c52a86a8
• chore(coding-agent): reference stable date format fix (fixes #2814)
  Mario Zechner · 2026-04-17 01:22:52 +02:00
  7f55605aa2
• fix(coding-agent): use stable date format in system prompt
  Mario Zechner · 2026-04-17 01:21:38 +02:00
  f81acc6679
• fix(coding-agent): omit devDependencies when installing packages (fixes #3009)
  Mario Zechner · 2026-04-17 01:21:31 +02:00
  ef1fcfcec2
• fix(coding-agent): restore /import error flow and document throws
  Mario Zechner · 2026-04-17 00:59:53 +02:00
  36bffc1d13
• fix(coding-agent): restore assistant/user turn spacing
  Mario Zechner · 2026-04-17 00:52:14 +02:00
  3ba0d85fa2
• fix(coding-agent): honor PI_PACKAGE_DIR for bun asset paths closes #3074
  Mario Zechner · 2026-04-17 00:38:34 +02:00
  e5f86f25f2
• test(ai): align adaptive thinking payload assertions
  Mario Zechner · 2026-04-17 00:35:07 +02:00
  8404fbafdb
• Merge branch 'main' of https://github.com/badlogic/pi-mono

  # Conflicts:
  #	packages/coding-agent/CHANGELOG.md

  Mario Zechner · 2026-04-17 00:28:14 +02:00
  b2ed851e84
• fix(coding-agent): stabilize user message OSC 133 padding closes #3090
  Mario Zechner · 2026-04-17 00:22:39 +02:00
  b0490310c3