What's Changed
- fix(models): frontier model check matches the model name only, never the provider route by @0xallam in #1280
- feat(reporting): link HTTP exchange evidence by @bearsyankees in #1281
- feat(prompts): require http_exchange_ids for proxy-validated findings by @0xallam in #1292
- docs: add Vercel AI Gateway provider guide by @erulkey in #1254
- fix(web_search): send only the agent's query to Exa search by @0xallam in #1306
- chore(deps): require litellm>=1.101.0 so gpt-6-astra accepts max_tokens by @0xallam in #1318
- feat(config): STRIX_API_TYPE forces responses vs chat completions by @bearsyankees in #1324
- fix(build): keep the TUI sidecar hook importable on hatchling 1.32.1 by @bearsyankees in #1325
- fix(runtime): place extra files as agent-writable sandbox files on every backend by @0xallam in #1330
- Prompt agents to include local Git blame in technical details by @bearsyankees in #1329
- fix(reporting): reduce git blame guidance to a trailing hint on the reporting tool by @0xallam in #1336
- runtime: read_only local sources become :ro bind mounts by @0xallam in #1342
- feat(mcp): initialize connections lazily by @yoni-at-strix in #1347
- feat(llm): structured per-attempt provider request log with provider request ids by @0xallam in #1353
- Let agents delete a vulnerability report they filed by @bearsyankees in #1354
- Fill in blank tool-call ids so strict providers accept the history by @0xallam in #1355
- fix(dev): make check-all non-mutating by @Czech-Knight in #1360
- fix(tui): suspend on ctrl+z by @ian-at-strix in #1371
- docs(skills): refresh framework behavior and security testing guidance by @bearsyankees in #1372
- perf(prompt): put per-run scope at the end of the system prompt by @ian-at-strix in #1375
- perf(llm): give Claude a cache point before the per-run scope by @ian-at-strix in #1376
- perf(prompt): load requested skills after a cache point by @ian-at-strix in #1382
- feat(tui): animate the wait_for_agents indicator by @ian-at-strix in #1383
- feat(budget): budget_policy=pause — park every agent at the cost limit until the operator resumes by @0xallam in #1387
- perf(llm): pin OpenRouter agents to one provider with session IDs by @ian-at-strix in #1386
- chore(models): remove the model quality warning and its allowlists by @0xallam in #1388
- fix(reporting): restore create_vulnerability_report parameter descrip… by @bearsyankees in #1391
- fix(config): choose Responses vs chat completions from the model, not the base URL by @0xallam in #1407
- fix(config): use the Responses API whenever the model supports it by @0xallam in #1408
- fix(finish_scan): stop asking for a section heading in every report field by @0xallam in #1411
- fix(tui): link every wrapped line of the viewer URL to the full URL by @0xallam in #1412
- feat(cli): Add --fail-on to gate headless exit code on severity by @siundu254 in #1399
- fix(cli): force UTF-8 stdout/stderr on Windows so Rich output never raises by @0xallam in #1414
- fix(preflight): name a non-ASCII character in the API key instead of raising UnicodeEncodeError by @0xallam in #1415
- fix(tui): render report section bodies as markdown by @0xallam in #1416
- fix(tui): collapsible, zoomable sidebar panels and a sidebar toggle by @0xallam in #1417
- fix(cli): verify the model before the TUI opens on a direct launch by @0xallam in #1418
- fix(telemetry): keep "Exception ignored" finalizer tracebacks off the terminal by @0xallam in #1419
- feat(cli): pick a prior run interactively when --resume has no name by @0xallam in #1421
- test(pricing): accept any identically priced provider for bare grok-4.5 by @0xallam in #1423
- fix(resume-picker): treat ctrl+c as cancel instead of leaking a KeyboardInterrupt traceback by @0xallam in #1424
- ci: run ruff, mypy, bandit, pytest, Go TUI and viewer checks on every pull request by @0xallam in #1422
- fix(deps): ship google-auth with every install so Vertex AI works in release binaries by @0xallam in #1437
- fix(preflight): give the startup model check its own 30s timeout instead of LLM_TIMEOUT by @0xallam in #1438
- fix(cli): run headless when no terminal is attached instead of crashing the TUI by @0xallam in #1440
- fix(docker): connect like the docker CLI and explain why the daemon is unreachable by @0xallam in #1441
- refactor(agents): replace respond_to_user with a text-free wait_for_user by @0xallam in #1442
- prompts: let the verified user define and change scope in chat by @0xallam in #1443
- readme: remove the Ask DeepWiki badge by @0xallam in #1444
- chore: release v1.7.0 by @0xallam in #1445
New Contributors
- @erulkey made their first contribution in #1254
- @Czech-Knight made their first contribution in #1360
- @ian-at-strix made their first contribution in #1371
- @siundu254 made their first contribution in #1399
Full Changelog: v1.6.2...v1.7.0