Adam Weidman
d5c0212b01
# PR Description: Refactor read_file Tool to Discourage Pagination
...
## Changes
- Removed `offset` and `limit` parameters from the `read_file` tool and the underlying `processSingleFileContent` utility.
- Updated `read_file` to always return the first 2000 lines of a text file if it exceeds that length.
- Added explicit guidance in the tool's truncation message and description, distinguishing between `grep_search` for pattern finding and `run_shell_command` (specifically `sed`) for extracting specific line ranges.
- Simplified `ProcessedFileReadResult` by removing `linesShown`.
- Updated all internal callers (including `ReadManyFilesTool`, `planCommand`, and `ExitPlanModeDialog`) to match the new signature of `processSingleFileContent`.
- Updated unit tests in `read-file.test.ts` and `fileUtils.test.ts` to verify the new 2000-line default truncation and remove pagination-related test cases.
## Motivation & Feedback
These changes were made to address feedback regarding how the model interacts with large files:
- **Discouraging Inefficient Pagination:** Feedback noted that providing `offset` and `limit` encouraged models to "paginate" through files chunk-by-chunk, which is often an inefficient "looping" behavior.
- **Encouraging Targeted Search:** By removing pagination parameters, we push the model toward more powerful and standard search behaviors. As the feedback stated: *"In my experience it's better to encourage the model to use grep or more complex bash commands to find what it needs."*
- **Compatibility with Search Results:** Models frequently use `grep -n` to find points of interest. Standard Unix tools like `sed` and `grep` are more compatible with 1-based line numbers and sophisticated searching than a manual 0-based `offset`/`limit` implementation.
- **Clarity on Truncation:** Addressed the feedback *"Could be useful to specify how will it be truncated"* by explicitly stating the 2000-line limit in the tool's output and description.
2026-02-09 13:56:02 -05:00
N. Taylor Mullen
fe70052baf
fix(evals): update save_memory evals and simplify tool description ( #18610 )
2026-02-09 01:06:03 -08:00
Andrew Garrett
8cbe851339
Fix newline insertion bug in replace tool ( #18595 )
2026-02-09 06:37:53 +00:00
N. Taylor Mullen
d45a45d565
chore: strengthen validation guidance in system prompt ( #18544 )
2026-02-09 05:32:46 +00:00
N. Taylor Mullen
cb73fbf384
feat(core): transition sub-agents to XML format and improve definitions ( #18555 )
2026-02-09 02:25:04 +00:00
Jerop Kipruto
375c104b32
fix(core): ensure enter_plan_mode tool registration respects experimental.plan ( #18587 )
2026-02-09 01:57:01 +00:00
N. Taylor Mullen
97a4e62dfa
feat(core): conditionally include ctrl+f prompt based on interactive shell setting ( #18561 )
2026-02-09 00:23:22 +00:00
Tommaso Sciortino
29a6aecffc
propagate abortSignal ( #18477 )
2026-02-08 21:56:26 +00:00
N. Taylor Mullen
92012365ca
fix(core): correct escaped interpolation in system prompt ( #18557 )
2026-02-08 21:08:17 +00:00
Abhi
11951592aa
feat(core): pause agent timeout budget while waiting for tool confirmation ( #18415 )
2026-02-08 04:03:47 +00:00
Abhi
bc8ffa6631
feat(context): add remote configuration for tool output masking thresholds ( #18553 )
2026-02-08 03:04:46 +00:00
N. Taylor Mullen
86bd7dbd4f
chore: remove feedback instruction from system prompt ( #18560 )
2026-02-08 02:22:50 +00:00
N. Taylor Mullen
eee95c509d
refactor(core): remove memory tool instructions from Gemini 3 prompt ( #18559 )
2026-02-08 01:57:53 +00:00
Abhi
a1f2aacd1e
feat(core): exempt high-signal tools from output masking ( #18545 )
2026-02-07 23:56:45 +00:00
Jerop Kipruto
be6723ebcc
chore: remove redundant planning prompt from final shell ( #18528 )
2026-02-07 19:45:09 +00:00
N. Taylor Mullen
9178b31629
feat(core): overhaul system prompt for rigor, integrity, and intent alignment ( #17263 )
2026-02-07 03:13:07 +00:00
Sandy Tao
19dc40825e
fix(core): expand excludeTools with legacy aliases for renamed tools ( #18498 )
2026-02-07 00:49:25 +00:00
Jerop Kipruto
e3796d137a
fix(core): prevent subagent bypass in plan mode ( #18484 )
2026-02-06 22:55:00 +00:00
Sandy Tao
28805a4b2d
refactor: simplify tool output truncation to single config ( #18446 )
2026-02-06 21:41:19 +00:00
Abhi
63f7e30790
feat(core): implement persistence and resumption for masked tool outputs ( #18451 )
2026-02-06 21:22:22 +00:00
joshualitt
7a8d6f6095
bug(core): Fix bug when saving plans. ( #18465 )
2026-02-06 20:45:37 +00:00
Shreya Keshive
f062f56b43
feat(admin): apply MCP allowlist to extensions & gemini mcp list command ( #18442 )
2026-02-06 18:07:00 +00:00
Sehoon Shon
61d92c4a21
Remove previewFeatures and default to Gemini 3 ( #18414 )
2026-02-06 18:02:57 +00:00
christine betts
099fea2434
Add extension registry client ( #18396 )
2026-02-06 17:14:14 +00:00
joshualitt
6fb3b09003
Shorten temp directory ( #17901 )
2026-02-06 16:10:17 +00:00
Abhi
30354892b3
feat(core,cli): implement session-linked tool output storage and cleanup ( #18416 )
2026-02-06 06:36:42 +00:00
Abhi
8ec176e005
feat(context): implement observation masking for tool outputs ( #18389 )
2026-02-06 01:53:11 +00:00
Adib234
4ffc349c18
feat(plan): create a metric for execution of plans generated in plan mode ( #18236 )
2026-02-05 23:46:34 +00:00
Adib234
fe975da91e
feat(plan): implement support for MCP servers in Plan mode ( #18229 )
2026-02-05 21:37:28 +00:00
Philippe
2498114df6
fix: resolve infinite loop when using 'Modify with external editor' ( #17453 )
...
Co-authored-by: Jack Wotherspoon <jackwoth@google.com >
Co-authored-by: ehedlund <ehedlund@google.com >
2026-02-05 20:52:41 +00:00
Tommaso Sciortino
9ca7300c90
Fix permission check ( #18395 )
2026-02-05 20:19:17 +00:00
Jerop Kipruto
dc09b4988d
feat(plan): integrate planning artifacts and tools into primary workflows ( #18375 )
2026-02-05 20:07:33 +00:00
Adib234
bce57ca1af
feat(plan): fix invalid tool calls in plan mode ( #18352 )
2026-02-05 20:04:03 +00:00
Jerop Kipruto
6860556afe
feat(plan): add guidance on iterating on approved plans vs creating new plans ( #18346 )
2026-02-05 19:11:45 +00:00
Alisa
5b9ea35b63
Improving memory tool instructions and eval testing ( #18091 )
2026-02-05 18:07:47 +00:00
Jerop Kipruto
4a6e3eb646
feat(plan): support replace tool in plan mode to edit plans ( #18379 )
2026-02-05 17:51:35 +00:00
Tommaso Sciortino
e4c80e6382
fix: Windows Specific Agent Quality & System Prompt ( #18351 )
2026-02-05 17:50:12 +00:00
Peter Friese
1cae5ab158
fix(core): handle non-compliant mcpbridge responses from Xcode 26.3 ( #18376 )
2026-02-05 17:03:32 +00:00
Philippe
e3b8490edf
fix(core): add retry logic for transient SSL/TLS errors ( #17318 ) ( #18310 )
2026-02-05 15:47:35 +00:00
Grant McCloskey
a3af4a8cae
feat(skills): implement linking for agent skills ( #18295 )
2026-02-04 22:11:01 +00:00
Chris Coutinho
821355c429
fix(mcp): ensure MCP transport is closed to prevent memory leaks ( #18054 )
...
Co-authored-by: Jack Wotherspoon <jackwoth@google.com >
2026-02-04 21:00:41 +00:00
Abhi
3afc8f25e1
fix(hooks): remove unnecessary logging for hook registration ( #18332 )
2026-02-04 19:48:33 +00:00
Jerop Kipruto
650980af37
feat(core): add enter_plan_mode tool ( #18324 )
2026-02-04 18:57:41 +00:00
Jack Wotherspoon
b987e1780d
feat: increase ask_user label limit to 16 characters ( #18320 )
2026-02-04 17:50:01 +00:00
Adib234
0012d95848
feat(plan): implement plan slash command ( #17698 )
2026-02-04 17:01:43 +00:00
Abhi
fedc0c5d60
feat(core): remove hardcoded policy bypass for local subagents ( #18153 )
2026-02-04 15:20:36 +00:00
gemini-cli-robot
99e523a15f
chore(release): bump version to 0.29.0-nightly.20260203.71f46f116 ( #18243 )
2026-02-04 14:43:29 +00:00
Christian Gunderman
a0b6602d09
Fix issue where agent gets stuck at interactive commands. ( #18272 )
2026-02-04 07:02:09 +00:00
Abhi
b39cefe14e
feat(core): add default execution limits for subagents ( #18274 )
2026-02-04 06:28:00 +00:00
Sehoon Shon
94f4e5cc15
feat(core): enable getUserTierName in config ( #18265 )
2026-02-04 03:28:58 +00:00
