vultr/openclaw
1
0
Fork 0
mirror of https://github.com/openclaw/openclaw.git synced 2026-04-27 09:02:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat(plugins): add before_agent_reply hook (claiming pattern) (#20067)

Browse Source 
Merged via squash.

Prepared head SHA: e40dfbdfb9
Co-authored-by: JoshuaLelon <23615754+JoshuaLelon@users.noreply.github.com>
Co-authored-by: jalehman <550978+jalehman@users.noreply.github.com>
Reviewed-by: @jalehman
This commit is contained in:
Joshua Lelon Mitchell
2026-04-01 15:31:11 -05:00
committed by GitHub
parent 017bc5261c
commit 7cb323d84f
8 changed files with 390 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
docs/automation/hooks.md
View File

@@ -607,7 +607,7 @@ Compaction lifecycle hooks exposed through the plugin hook runner:
### Complete Plugin Hook Reference
All 27 hooks registered via the Plugin SDK. Hooks marked **sequential** run in priority order and can modify results; **parallel** hooks are fire-and-forget.
All 28 hooks registered via the Plugin SDK. Hooks marked **sequential** run in priority order and can modify results; **parallel** hooks are fire-and-forget.
#### Model and prompt hooks
@@ -616,6 +616,7 @@ All 27 hooks registered via the Plugin SDK. Hooks marked **sequential** run in p
| `before_model_resolve` | Before model/provider lookup | Sequential | `{ modelOverride?, providerOverride? }` |
| `before_prompt_build` | After model resolved, session messages ready | Sequential | `{ systemPrompt?, prependContext?, appendSystemContext? }` |
| `before_agent_start` | Legacy combined hook (prefer the two above) | Sequential | Union of both result shapes |
| `before_agent_reply` | After inline actions, before the LLM runs | Sequential | `{ handled: boolean, reply?, reason? }` |
| `llm_input` | Immediately before the LLM API call | Parallel | `void` |
| `llm_output` | Immediately after LLM response received | Parallel | `void` |

1
docs/concepts/agent-loop.md
View File

@@ -84,6 +84,7 @@ These run inside the agent loop or gateway pipeline:
- **`before_model_resolve`**: runs pre-session (no `messages`) to deterministically override provider/model before model resolution.
- **`before_prompt_build`**: runs after session load (with `messages`) to inject `prependContext`, `systemPrompt`, `prependSystemContext`, or `appendSystemContext` before prompt submission. Use `prependContext` for per-turn dynamic text and system-context fields for stable guidance that should sit in system prompt space.
- **`before_agent_start`**: legacy compatibility hook that may run in either phase; prefer the explicit hooks above.
- **`before_agent_reply`**: runs after inline actions and before the LLM call, letting a plugin claim the turn and return a synthetic reply or silence the turn entirely.
- **`agent_end`**: inspect the final message list and run metadata after completion.
- **`before_compaction` / `after_compaction`**: observe or annotate compaction cycles.
- **`before_tool_call` / `after_tool_call`**: intercept tool params/results.