chengyongru 93bcb0a649 fix(agent): prevent history.jsonl bloat from raw_archive and stuck consolidation ...

Root cause: when consolidation LLM fails, raw_archive() dumped full message
content (~1MB) into history.jsonl with no size limit. Since build_system_prompt()
injects history.jsonl into every system prompt, all subsequent LLM calls exceeded
the 200K context window with error 1261.

Additionally, _cap_consolidation_boundary's 60-message cap caused consolidation
to get stuck on sessions with long tool chains (200+ iterations), triggering
the raw_archive fallback in the first place.

Three-layer fix:
- Remove _cap_consolidation_boundary: let pick_consolidation_boundary drive
  chunk sizing based solely on token budget
- Truncate archive() input: use tiktoken to cap formatted text to the model's
  input token budget before sending to consolidation LLM
- Truncate raw_archive() output: cap history.jsonl entries at 16K chars

2026-04-24 01:41:01 +08:00

..

agent

fix(agent): prevent history.jsonl bloat from raw_archive and stuck consolidation

2026-04-24 01:41:01 +08:00

channels

fix(transcription): harden language parameter validation and tests

2026-04-22 12:41:32 +08:00

cli

fix(session): fsync sessions on graceful shutdown to prevent data loss

2026-04-22 13:19:53 +08:00

command

fix(commands): intercept non-priority commands during active turn

2026-04-21 21:50:37 +08:00

config

refactor(config): nest MyTool settings under tools.my (with legacy-key migration)

2026-04-16 15:58:20 +00:00

cron

fix(cron): drop top-level oneOf so OpenAI Codex/Responses accept tool schema

2026-04-19 21:54:38 +08:00

providers

review(providers): tighten comments in reasoning_effort normalize path

2026-04-22 12:49:55 +08:00

security

fix(security): add ssrfWhitelist config to unblock Tailscale/CGNAT (#2669)

2026-04-04 19:43:18 +08:00

session

fix: handle Windows PermissionError on directory fsync

2026-04-22 13:19:53 +08:00

tools

feat(read_file): add DOCX, XLSX, PPTX support via document.extract_text()

2026-04-21 22:12:19 +08:00

utils

fix(utils): handle malformed think tags and channel markers in strip_think

2026-04-20 17:04:48 +08:00

test_api_attachment.py

fix: two bugs in document extraction pipeline

2026-04-14 13:15:04 +00:00

test_api_stream.py

Prevent failed SSE requests from masquerading as successful completions

2026-04-18 00:44:44 +08:00

test_build_status.py

fix(status): correct context percentage calculation and sync consolidator

2026-04-16 20:30:39 +08:00

test_context_documents.py

fix: two bugs in document extraction pipeline

2026-04-14 13:15:04 +00:00

test_docker.sh

docs: add core agent line count script and update README with real-time stats

2026-02-06 07:28:39 +00:00

test_document_parsing.py

fix(utils): recurse into PPTX groups and tables when extracting text (#3250)

2026-04-18 12:30:42 +08:00

test_msteams.py

fix(msteams): secure inbound defaults and ref persistence

2026-04-16 13:22:07 +08:00

test_nanobot_facade.py

fix(sdk): route github copilot through oauth provider

2026-04-02 03:46:40 +08:00

test_openai_api.py

feat(api): add SSE streaming for /v1/chat/completions Wire up the existing on_stream/on_stream_end callbacks from process_direct() to emit OpenAI-compatible SSE chunks when stream=true. Non-streaming path is untouched.

2026-04-17 01:54:49 +08:00

test_package_version.py

fix(version): fall back to pyproject in source checkouts

2026-04-08 23:47:36 +08:00

test_tool_contextvars.py

test(tools): add basic regression tests for ContextVar routing context

2026-04-21 13:25:30 +08:00

test_truncate_text_shadowing.py

fix(agent): avoid truncate_text name shadowing

2026-04-10 17:36:31 +08:00