deepagents-talon==0.0.9

### Features

- Add a Slack channel adapter, with support for outbound user mentions. ([#6575](https://github.com/langchain-ai/deepagents/pull/6575), [#6602](https://github.com/langchain-ai/deepagents/pull/6602))
- Add sender pairing for DM access and Slack, including DM delivery of pairing codes for Slack mentions. Paired senders are admitted in every chat. ([#6579](https://github.com/langchain-ai/deepagents/pull/6579), [#6582](https://github.com/langchain-ai/deepagents/pull/6582), [#6584](https://github.com/langchain-ai/deepagents/pull/6584), [#6592](https://github.com/langchain-ai/deepagents/pull/6592))
- Add `/model` for per-chat model switching. ([#6577](https://github.com/langchain-ai/deepagents/pull/6577))
- Add cron expressions and `until` expiry for scheduled jobs, run their subagents inline, and deliver results from thread-based jobs to the channel by default. ([#6526](https://github.com/langchain-ai/deepagents/pull/6526), [#6425](https://github.com/langchain-ai/deepagents/pull/6425), [#6632](https://github.com/langchain-ai/deepagents/pull/6632))
- Add opt-in sandboxed execution via `DEEPAGENTS_TALON_SANDBOX`. ([#6574](https://github.com/langchain-ai/deepagents/pull/6574))
- Register chat commands as Discord slash commands and share public thread history by guild channel. ([#6303](https://github.com/langchain-ai/deepagents/pull/6303), [#6606](https://github.com/langchain-ai/deepagents/pull/6606))
- Add approved, one-off `ask_for_help` consultations. ([#6605](https://github.com/langchain-ai/deepagents/pull/6605))
- Batch approvals for concurrent tool calls. ([#6435](https://github.com/langchain-ai/deepagents/pull/6435))
- Add a `/context-doctor` command. ([#6557](https://github.com/langchain-ai/deepagents/pull/6557))
- Add a general-purpose safety preflight skill. ([#6587](https://github.com/langchain-ai/deepagents/pull/6587))
- Support explicitly configured public MCP OAuth clients. ([#6670](https://github.com/langchain-ai/deepagents/pull/6670))

### Bug Fixes

- Retry provider overload and connection errors, including overload errors without a status code. ([#6580](https://github.com/langchain-ai/deepagents/pull/6580), [#6304](https://github.com/langchain-ai/deepagents/pull/6304))
- Transcribe inbound audio attachments, preserve MIME types for downloaded WhatsApp voice messages, and avoid retranscribing voice on background follow-ups. ([#6680](https://github.com/langchain-ai/deepagents/pull/6680), [#6590](https://github.com/langchain-ai/deepagents/pull/6590), [#6682](https://github.com/langchain-ai/deepagents/pull/6682))
- List selectable gateway models and persist model selection across conversations. ([#6595](https://github.com/langchain-ai/deepagents/pull/6595), [#6603](https://github.com/langchain-ai/deepagents/pull/6603))
- Preserve Slack conversation context across threads, include preceding replies when mentioned in a thread, and keep legacy thread archives writable. ([#6665](https://github.com/langchain-ai/deepagents/pull/6665), [#6634](https://github.com/langchain-ai/deepagents/pull/6634), [#6679](https://github.com/langchain-ai/deepagents/pull/6679))
- Allow scheduled jobs to read their origin chat’s history and serialize changes to the cron store. ([#6583](https://github.com/langchain-ai/deepagents/pull/6583), [#6688](https://github.com/langchain-ai/deepagents/pull/6688))
- Accept any Slack slash command name and labeled Slack OAuth callbacks, while excluding OAuth callbacks from thread context. ([#6589](https://github.com/langchain-ai/deepagents/pull/6589), [#6593](https://github.com/langchain-ai/deepagents/pull/6593), [#6687](https://github.com/langchain-ai/deepagents/pull/6687))
- Deliver narration before running tools and preserve turn cancellation during typing cleanup. ([#6681](https://github.com/langchain-ai/deepagents/pull/6681), [#6594](https://github.com/langchain-ai/deepagents/pull/6594))
- Index only visible conversation text and persist history-vector deduplication. ([#6358](https://github.com/langchain-ai/deepagents/pull/6358), [#6357](https://github.com/langchain-ai/deepagents/pull/6357))
- Improve subagent research handoffs. ([#6674](https://github.com/langchain-ai/deepagents/pull/6674))
- Store offloaded artifacts in the assistant home. ([#6530](https://github.com/langchain-ai/deepagents/pull/6530))
- Validate archive pagination arguments. ([#6442](https://github.com/langchain-ai/deepagents/pull/6442))
- Declare packages that Talon imports directly as dependencies. ([#6436](https://github.com/langchain-ai/deepagents/pull/6436))

### Performance Improvements

- Reduce transactions when replaying conversation archives. ([#6320](https://github.com/langchain-ai/deepagents/pull/6320))

---

Thanks to our community contributors: @akyourowngames

**Special thanks** to everyone who reported the issues addressed in this release: @akyourowngames ([#6335](https://github.com/langchain-ai/deepagents/issues/6335))

Internal maintainers: @jkennedyvz, @linted

Released by: @jkennedyvz

<details>
<summary>Git log since deepagents-talon==0.0.8</summary>

This commit history includes changes to this package. Commits are listed newest first.

- [`3610d29`](https://github.com/langchain-ai/deepagents/commit/3610d29e4174c677e69366874498a7ace005f252) release(deepagents-talon): 0.0.9 (#6305)
- [`f566a09`](https://github.com/langchain-ai/deepagents/commit/f566a09da6141935ee564caa01c921da08f34036) fix(talon): exclude OAuth callbacks from Slack thread context (#6687)
- [`67b0bef`](https://github.com/langchain-ai/deepagents/commit/67b0bef01082bf94e8641bf4e1a94a71b3f6ebc9) fix(talon): serialize cron store mutations (#6688)
- [`1614a38`](https://github.com/langchain-ai/deepagents/commit/1614a389ea48c12182c167d3ecf69224a69d4043) fix(talon): avoid retranscribing voice on background follow-ups (#6682)
- [`a8bf07b`](https://github.com/langchain-ai/deepagents/commit/a8bf07bf604524e98530034027414254ae215f78) fix(talon): deliver narration before running tools (#6681)
- [`184f547`](https://github.com/langchain-ai/deepagents/commit/184f547e12f57df358eb49536a02cbe672f95888) fix(talon): transcribe inbound audio attachments (#6680)
- [`7475059`](https://github.com/langchain-ai/deepagents/commit/7475059432e9eda77b8b9a5bc7178f5f3a532bd9) feat(talon): support explicit public MCP OAuth clients (#6670)
- [`166b0e0`](https://github.com/langchain-ai/deepagents/commit/166b0e0ce1e424bb24f49ee5c48155458b4b1dce) fix(talon): keep legacy Slack thread archives writable (#6679)
- [`4394bcd`](https://github.com/langchain-ai/deepagents/commit/4394bcd00b8eb46e7c423939643a0dfcfb5d8773) release(deepagents): 0.7.21 (#6676)
- [`ab57eab`](https://github.com/langchain-ai/deepagents/commit/ab57eab7e51121a2677d0c5b5aa396487ed2b9ee) fix(talon): improve subagent research handoffs (#6674)
- [`6bf7665`](https://github.com/langchain-ai/deepagents/commit/6bf76658f88af8328d9a9625d2c6444e01896248) fix(talon): share Slack channel conversation archives across threads (#6665)
- [`27196b5`](https://github.com/langchain-ai/deepagents/commit/27196b5f8b299c178a54e30b280a2c05c19258c2) build(deps): bump brace-expansion from 2.1.4 to 2.1.7 in /libs/talon/deepagents_talon/channels/whatsapp_bridge (#6669)
- [`97cd00e`](https://github.com/langchain-ai/deepagents/commit/97cd00ee1340a92982adf2731352252288d76755) build(deps): bump pyjwt from 2.15.0 to 2.15.1 in /libs/talon (#6668)
- [`91eda86`](https://github.com/langchain-ai/deepagents/commit/91eda86832d2e3a9c0d14ca54da3f07be4ff37ce) build(deps): bump pyjwt from 2.13.0 to 2.15.0 in /libs/talon (#6646)
- [`6639d12`](https://github.com/langchain-ai/deepagents/commit/6639d1262eeb5ab857de015689238ba5b2b5d51e) build(deps): bump urllib3 from 2.7.0 to 2.8.0 in /libs/talon (#6647)
- [`0e5f6f8`](https://github.com/langchain-ai/deepagents/commit/0e5f6f893b91bed16837f2cd7966049c7cafc650) fix(talon): include preceding Slack replies on thread mentions (#6634)
- [`c9b2ce1`](https://github.com/langchain-ai/deepagents/commit/c9b2ce194e422d2a61b9faf3732da15d90623c13) chore(deps): raise dependency minimums for `all` (#6645)
- [`f6f9721`](https://github.com/langchain-ai/deepagents/commit/f6f97216bae2d1c969fe6c045420b4dd75795bdb) release(deepagents-code): 0.1.79 (#6615)
- [`07628a5`](https://github.com/langchain-ai/deepagents/commit/07628a59010348aca443087a19c477f148997b5b) feat(talon): deliver thread cron results to the channel by default (#6632)
- [`ff7cb5c`](https://github.com/langchain-ai/deepagents/commit/ff7cb5cbf7c6d2b8804d25283deb6dfc2dc998f1) refactor(talon): remove legacy cron job compatibility (#6620)
- [`6bf7802`](https://github.com/langchain-ai/deepagents/commit/6bf780284ec49789613df20d85866a0ad981e579) release(deepagents): 0.7.20 (#6624)
- [`a87e219`](https://github.com/langchain-ai/deepagents/commit/a87e21955cb42c47f1e2ee728a37b76547939567) chore(deps): raise dependency minimums for `all` (#6631)
- [`433784e`](https://github.com/langchain-ai/deepagents/commit/433784ee99db01a1c98c058955c41e6a03462d2d) release(langchain-quickjs): 0.3.8 (#6629)
- [`689898e`](https://github.com/langchain-ai/deepagents/commit/689898e50a0cab7e05b5b81132b7cee4304e0020) chore(deps): raise dependency minimums for `all` (#6628)
- [`3175d23`](https://github.com/langchain-ai/deepagents/commit/3175d231f192db63cbb432c6ca133d9486dc59ba) refactor(talon): simplify startup and test behavior instead of structure (#6619)
- [`e895ae6`](https://github.com/langchain-ai/deepagents/commit/e895ae6810fae010d56cef6ecfcaccbc29632b25) refactor(talon): remove legacy send result normalization (#6618)
- [`019c112`](https://github.com/langchain-ai/deepagents/commit/019c1120608c0838f5c539e2c6fc7696e7cfeb50) test(talon): remove vestigial Fleet configuration tests (#6617)
- [`936a37e`](https://github.com/langchain-ai/deepagents/commit/936a37ef7b85a5ea3a2dca989e3e3a218fb96700) feat(talon): share public Discord thread history by guild channel (#6606)
- [`15d8f47`](https://github.com/langchain-ai/deepagents/commit/15d8f471d8c6ba6868fc6d94beb7fad89ef7406f) feat(talon): add approved one-off `ask_for_help` consultations (#6605)
- [`eff90de`](https://github.com/langchain-ai/deepagents/commit/eff90dec8c6bfe9636da064c7c9310b660d89b32) fix(talon): stop swallowing a turn's cancel during typing cleanup (#6594)
- [`156a682`](https://github.com/langchain-ai/deepagents/commit/156a6824e53c3f483dca75d08e0d084c0788b3a3) fix(talon): persist model selection across conversations (#6603)
- [`b329965`](https://github.com/langchain-ai/deepagents/commit/b32996518f9bad4d156d19e35915d51ead7e6019) feat(talon): allow outbound Slack user mentions (#6602)
- [`8b2a526`](https://github.com/langchain-ai/deepagents/commit/8b2a5267ab1fdcf523bac6c3d705f00111869f9a) release(deepagents-code): 0.1.78 (#6600)
- [`5b29445`](https://github.com/langchain-ai/deepagents/commit/5b29445dd387bb9e8b8c3fed8f7382a91a990e06) fix(talon): enumerate selectable gateway models (#6595)
- [`e739e4d`](https://github.com/langchain-ai/deepagents/commit/e739e4d66c367f91b430062ee841055ac311c57c) fix(talon): preserve MIME types when downloading WhatsApp voice messages (#6590)
- [`512334f`](https://github.com/langchain-ai/deepagents/commit/512334f09b21e2bb4cf4b039ef5f89752ed42026) chore(deps): raise dependency minimums for `all` (#6548)
- [`b3f1a39`](https://github.com/langchain-ai/deepagents/commit/b3f1a3919fe39853ee29acee1fb49b077507428f) fix(talon): retry provider overload and connection errors (#6580)
- [`6744083`](https://github.com/langchain-ai/deepagents/commit/6744083d1978b187fce1de0226aad3c5c9a647dc) feat(talon): admit paired senders in every chat (#6592)
- [`0359577`](https://github.com/langchain-ai/deepagents/commit/035957706fc34545ae1e7485aa03f8415694cd71) fix(talon): accept labeled Slack OAuth callbacks (#6593)
- [`86923e1`](https://github.com/langchain-ai/deepagents/commit/86923e160c18a6c090045ef1d60a17bd0d65760d) feat(talon): add general-purpose safety preflight skill (#6587)
- [`31ae532`](https://github.com/langchain-ai/deepagents/commit/31ae53291c23854912d3237c20e108445c6b6f11) fix(talon): accept any Slack slash command name (#6589)
- [`a31b310`](https://github.com/langchain-ai/deepagents/commit/a31b31009b3f7fad90cc567cd794def20e280bd3) feat(talon): send Slack mention pairing codes in DMs (#6584)
- [`e70cf09`](https://github.com/langchain-ai/deepagents/commit/e70cf09267d93ce54f381d77fb577a76adac6300) fix(talon): let scheduled jobs read their origin chat's history (#6583)
- [`90cd463`](https://github.com/langchain-ai/deepagents/commit/90cd463830418f79dc994f7b84c01f6256409435) feat(talon): support sender pairing on Slack (#6582)
- [`40873ba`](https://github.com/langchain-ai/deepagents/commit/40873baa880133ae19ea921239ad9142c1972f4f) feat(talon): add `/model` command for per-chat model switching (#6577)
- [`135e849`](https://github.com/langchain-ai/deepagents/commit/135e849484d15974a5dbc997828c9889b2d098fd) feat(talon): add sender pairing for DM access (#6579)
- [`b5f22b0`](https://github.com/langchain-ai/deepagents/commit/b5f22b01aff7cfe328d632de086c31247c2b1081) feat(talon): opt-in sandboxed execution via `DEEPAGENTS_TALON_SANDBOX` (#6574)
- [`15bd7fd`](https://github.com/langchain-ai/deepagents/commit/15bd7fdcfb5b6a02a650a5faa7aced0faf6dd6d8) feat(talon): add Slack channel adapter (#6575)
- [`71ae3f6`](https://github.com/langchain-ai/deepagents/commit/71ae3f69b161fdb0d009fe91e112e9aa39369a4d) feat(talon): add cron expressions and `until` expiry to scheduled jobs (#6526)
- [`60c0228`](https://github.com/langchain-ai/deepagents/commit/60c0228aad14d042080743ead04e135994d3fa75) feat(talon): add `/context-doctor` command (#6557)
- [`dcf4feb`](https://github.com/langchain-ai/deepagents/commit/dcf4febea597a876043f0aea6ecae40c321ea541) release(deepagents-code): 0.1.77 (#6545)
- [`bb49cf5`](https://github.com/langchain-ai/deepagents/commit/bb49cf50d3450a3bbebe8d25240cb7ef6379bd0d) release(deepagents): 0.7.19 (#6523)
- [`e801fa1`](https://github.com/langchain-ai/deepagents/commit/e801fa1810f971e693b8965b5b9802803aeb555d) release(deepagents-code): 0.1.76 (#6538)
- [`123f9c3`](https://github.com/langchain-ai/deepagents/commit/123f9c3030548e609798e54c11ca9455653b5764) chore(deps): raise dependency minimums for `all` (#6535)
- [`74d768b`](https://github.com/langchain-ai/deepagents/commit/74d768bd45dd4eca977dc320ebfe5838f1691682) fix(talon): store offloaded artifacts in the assistant home (#6530)
- [`c526854`](https://github.com/langchain-ai/deepagents/commit/c52685431535b143b860a995eaadfbabcf9b409c) release(deepagents-code): 0.1.75 (#6518)
- [`460feb6`](https://github.com/langchain-ai/deepagents/commit/460feb63301e61e7e40e27932ca41a908c442841) chore(deps): raise dependency minimums for `all` (#6511)
- [`e109d04`](https://github.com/langchain-ai/deepagents/commit/e109d04a9a61cfaf7719211c1e3c35889127eae2) release(deepagents-code): 0.1.74 (#6493)
- [`c172018`](https://github.com/langchain-ai/deepagents/commit/c17201847dd65390662a763ddd93ed84bbb8931c) release(deepagents): 0.7.18 (#6491)
- [`9a50bf5`](https://github.com/langchain-ai/deepagents/commit/9a50bf5c966fb599e240a6c52167133daaff5565) chore(deps): raise dependency minimums for `all` (#6498)
- [`d5a1896`](https://github.com/langchain-ai/deepagents/commit/d5a18968f2669ce08290b2d7e2c46da13b696efa) release(deepagents-code): 0.1.73 (#6457)
- [`2b767bf`](https://github.com/langchain-ai/deepagents/commit/2b767bfb8628ec50fc775fef85cbe780d2d22451) chore(deps): raise dependency minimums for `all` (#6480)
- [`e5406d0`](https://github.com/langchain-ai/deepagents/commit/e5406d0c735c7eb72548b7ea616991987bfb21e6) release(deepagents): 0.7.17 (#6458)
- [`cf5c560`](https://github.com/langchain-ai/deepagents/commit/cf5c5606648efa8bb27684fac97d9e794039707d) chore(deps): raise dependency minimums for `all` (#6478)
- [`0114a3f`](https://github.com/langchain-ai/deepagents/commit/0114a3f824cd5e626acb7b64d3c8e24468319d09) release(deepagents): 0.7.16 (#6372)
- [`a764619`](https://github.com/langchain-ai/deepagents/commit/a764619aa8c850bc75e2e916cf53a587637d8c81) release(deepagents-code): 0.1.72 (#6403)
- [`f173a01`](https://github.com/langchain-ai/deepagents/commit/f173a011a60d73163410f88d3b6cbf33670e768c) chore(deps): raise dependency minimums for `all` (#6440)
- [`f84d963`](https://github.com/langchain-ai/deepagents/commit/f84d96358a111872fba3f9b69c5b08ad1565c083) fix(talon): validate archive pagination arguments (#6442)
- [`1660147`](https://github.com/langchain-ai/deepagents/commit/16601472424400ab4a90608d7d95e8fc61f23267) feat(talon): batch concurrent tool approvals (#6435)
- [`c3a041e`](https://github.com/langchain-ai/deepagents/commit/c3a041e3d8f593e4e4c9bfc273d52264ceff165b) feat(talon): run a scheduled job's subagents inline (#6425)
- [`c63eba4`](https://github.com/langchain-ai/deepagents/commit/c63eba426b28f1d1685e40767f087dcc3f34334c) fix(talon): declare the packages talon imports directly (#6436)
- [`0923301`](https://github.com/langchain-ai/deepagents/commit/092330158bab829cab80d930b05b2b4c512f8f38) refactor(talon): move MCP onto `langchain.mcp` (#6091)
- [`aff6d16`](https://github.com/langchain-ai/deepagents/commit/aff6d168fcc0c2adab2666abf5dde6de4ca421a9) docs(talon): point WhatsApp example link at the real directory (#6354)
- [`f6850e9`](https://github.com/langchain-ai/deepagents/commit/f6850e9d25854b5679010c08f02adae99e6b8a2c) fix(talon): index only visible conversation text (#6358)
- [`b3d4967`](https://github.com/langchain-ai/deepagents/commit/b3d4967bdcda5b42e976b3e97c357deea82d63ac) fix(talon): persist history vector deduplication (#6357)
- [`229ffef`](https://github.com/langchain-ai/deepagents/commit/229ffef3d41bbe6dcfff3852bbd2058331aabe01) release(deepagents-code): 0.1.70 (#6323)
- [`0f5a2b5`](https://github.com/langchain-ai/deepagents/commit/0f5a2b57fa5dbb3a7d8f16dc280cb5b1506ea8c0) release(deepagents): 0.7.15 (#6298)
- [`7a7f382`](https://github.com/langchain-ai/deepagents/commit/7a7f3827430572a0976002003d9f19d48ae20360) chore(deps): raise dependency minimums for `all` (#6336)
- [`06e5886`](https://github.com/langchain-ai/deepagents/commit/06e588610c08e7544ee77ac09aa971921a5c893e) chore(deps): raise dependency minimums for `all` (#6310)
- [`89bd48f`](https://github.com/langchain-ai/deepagents/commit/89bd48f7d531c9affafa3f0dc096083cbc5969dd) perf(talon): reduce archive replay transactions (#6320)
- [`1d3232c`](https://github.com/langchain-ai/deepagents/commit/1d3232c0852c47af09119edea10eeec887e4f0da) release(deepagents-code): 0.1.69 (#6247)
- [`f36b641`](https://github.com/langchain-ai/deepagents/commit/f36b641ce3e1f44dbd47835d6dc9f5bb1c82324a) feat(talon): register chat commands as Discord slash commands (#6303)
- [`f8acbd0`](https://github.com/langchain-ai/deepagents/commit/f8acbd0ae24278d35aeb1fa014056084a58f70d2) fix(talon): retry statusless provider overload errors (#6304)
- [`ded0f11`](https://github.com/langchain-ai/deepagents/commit/ded0f1187f6bf81e2f679ea7b16ead7760d775a6) release(deepagents): 0.7.14 (#6050)
- [`3b99e73`](https://github.com/langchain-ai/deepagents/commit/3b99e7333e501d1ccf84b5fbaaacc9de815ce4b7) chore(deps): raise dependency minimums for `all` (#6280)
- [`9f6bd6a`](https://github.com/langchain-ai/deepagents/commit/9f6bd6ae200ff2e45e8c0246343b6c8936707447) chore(deps): raise dependency minimums for `all` (#6268)

</details>