Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
167 commits
Select commit Hold shift + click to select a range
d65dc4e
add dummy tools
eb8680 Feb 8, 2026
fa3fb6b
Add Agent class
eb8680 Feb 8, 2026
eca5cb8
stash examples
eb8680 Feb 8, 2026
310e3db
no changes to library code
eb8680 Apr 25, 2026
136b3a8
no agent test
eb8680 Apr 25, 2026
3b32604
Merge branch 'master' into eb-llm-examples
eb8680 Apr 25, 2026
52dc30e
updates
eb8680 Apr 25, 2026
2cbb0e0
notebook sections as scripts
eb8680 Apr 25, 2026
ed3d16b
lint
eb8680 Apr 25, 2026
3cac33d
nit
eb8680 Apr 25, 2026
ee6fc1f
retry
eb8680 Apr 25, 2026
c2f6bbc
Add example using object handles (#597)
jfeser Jun 2, 2026
c3555c4
Synthesis-based template handling
eb8680 Jun 19, 2026
1eaa685
doctest
eb8680 Jun 19, 2026
d5129c8
Add idiomatic codeadapt example (#695)
eb8680 Jun 20, 2026
70afc43
observability
eb8680 Jun 22, 2026
598c687
Support bound method Templates
eb8680 Jun 23, 2026
ad2e585
cleanup
eb8680 Jun 24, 2026
ce03b30
add name and signature to prompt_template and tool
eb8680 Jun 29, 2026
9c775f3
add serialization-only schema for callable and add return schema to t…
eb8680 Jun 29, 2026
00afd96
Assemble system prompt
eb8680 Jun 30, 2026
e17af14
many simplifications
eb8680 Jul 1, 2026
eecd8ec
Add terminal renderer
eb8680 Jul 1, 2026
991be62
rename system prompt block helpers to avoid renderer confusion
eb8680 Jul 1, 2026
cc0b742
Tweak system prompt
eb8680 Jul 1, 2026
d515a41
Remove IsRecursive
eb8680 Jul 1, 2026
0ba7274
include docstring
eb8680 Jul 1, 2026
1a1f42f
move around
eb8680 Jul 2, 2026
f5d0440
Merge branch 'master' into eb-doctest-synthesis
eb8680 Jul 11, 2026
c26081f
formatting nits
eb8680 Jul 13, 2026
600f058
formatting nits
eb8680 Jul 13, 2026
a44624b
Merge branch 'master' into eb-doctest-synthesis
eb8680 Jul 15, 2026
8c11615
fix failed merge
eb8680 Jul 18, 2026
6b665aa
split into more files
eb8680 Jul 19, 2026
395c51f
Add a library-level harness.py
eb8680 Jul 19, 2026
3b0a01f
Merge branch 'eb-llm-examples' into eb-doctest-synthesis
eb8680 Jul 19, 2026
a3cf92b
standardize harness
eb8680 Jul 19, 2026
b0fd2cd
cleanup examples
eb8680 Jul 19, 2026
234c059
nits
eb8680 Jul 19, 2026
9902b84
fix docstring
eb8680 Jul 19, 2026
2cc4c35
tool choice in harness
eb8680 Jul 19, 2026
d0f0d4b
contextvar
eb8680 Jul 19, 2026
ad8afa8
remove dumb example
eb8680 Jul 19, 2026
bcb5a0c
image
eb8680 Jul 19, 2026
1c5cdc6
fix synthesisfinaltool tool collection bug
eb8680 Jul 20, 2026
969c703
remove hanoi hints
eb8680 Jul 20, 2026
e6f5c71
fix bug
eb8680 Jul 20, 2026
6e64028
nit
eb8680 Jul 20, 2026
ed6de7d
comment about nasty bug
eb8680 Jul 20, 2026
bb67580
Move Encodable to template.py and fix lint
eb8680 Jul 20, 2026
8c7b9b0
nit
eb8680 Jul 20, 2026
4cf864e
revert
eb8680 Jul 20, 2026
958415d
minor
eb8680 Jul 20, 2026
44c251a
Merge branch 'master' into eb-doctest-synthesis
eb8680 Jul 20, 2026
22bd549
repl_history
eb8680 Jul 20, 2026
b244167
Merge branch 'master' into eb-doctest-synthesis
eb8680 Jul 23, 2026
a623637
cleanup
eb8680 Jul 23, 2026
62ebd45
force type checking of submit_solution
eb8680 Jul 23, 2026
fda9049
remove trivial majority_vote
eb8680 Jul 23, 2026
867939d
many fixes
eb8680 Jul 24, 2026
7ecf823
fix no-tool-call error
eb8680 Jul 24, 2026
998c59a
fix prompts
eb8680 Jul 24, 2026
2537af0
fix bug in method templates that synthesize code
eb8680 Jul 24, 2026
4717f2f
fix test bugs
eb8680 Jul 24, 2026
7f9476a
add schema-harness example in world_model_agent.py
eb8680 Jul 24, 2026
a348f8a
add pdb and reasoning-effort options to harness
eb8680 Jul 24, 2026
f349948
help = doc
eb8680 Jul 24, 2026
7ba88ac
explore
eb8680 Jul 24, 2026
9763421
consolidate hanoi
eb8680 Jul 24, 2026
b7d859a
add a summary comment
eb8680 Jul 24, 2026
53dea20
reorganize examples
eb8680 Jul 25, 2026
c944e7d
remove bad example
eb8680 Jul 25, 2026
587fe8a
format
eb8680 Jul 25, 2026
96edcbe
nit
eb8680 Jul 25, 2026
a44f5d8
Merge branch 'master' into eb-doctest-synthesis
eb8680 Jul 26, 2026
23cb7e4
fix nested agent history
eb8680 Jul 26, 2026
61be1b6
port persistent agent (#604)
eb8680 Jul 27, 2026
86d1c51
add more aime examples and consolidate into one file
eb8680 Jul 28, 2026
6730683
Port multi-agent choreography (#607) to asyncio
eb8680 Jul 28, 2026
cb069f1
Move multi-agent example under llm_examples
eb8680 Jul 28, 2026
cca3fb2
Replace the TaskQueue backends with asyncio futures and a Queue
eb8680 Jul 28, 2026
0f23524
Rename multi.py to choreographies.py
eb8680 Jul 28, 2026
2000219
Add StepLog, a write-ahead log of completed steps, for resume
eb8680 Jul 28, 2026
d05cf18
Give reviewers a run_tests tool, so the fix loop turns on a fact
eb8680 Jul 28, 2026
bf500e2
Drop Choreography's handlers argument
eb8680 Jul 28, 2026
8569080
Make step, call and scatter Operations handled by EndpointProjection
eb8680 Jul 28, 2026
b3fd65a
Fold call into step, and inline _first_exception and replay
eb8680 Jul 28, 2026
8635b9d
Remove Choreography.projection
eb8680 Jul 28, 2026
8b2be03
Absorb StepLog into _Steps; Choreography takes a log path
eb8680 Jul 28, 2026
8e154c2
Drop recorded_steps; fold _connect into _Steps; trim the docstring
eb8680 Jul 28, 2026
d894aac
Make Choreography generic in its program's signature, and callable
eb8680 Jul 28, 2026
622979a
Give run_doctests a default rule, and make RestrictedEvalProvider usable
eb8680 Jul 28, 2026
6094ecc
Let the harness choose which provider runs model-authored Python
eb8680 Jul 28, 2026
4539b77
Stop re-validating a callable against synthesis constraints when enco…
eb8680 Jul 28, 2026
c728275
Fix two examples that could not run at all
eb8680 Jul 28, 2026
1335319
Merge branch 'eb-doctest-synthesis' into worktree-restricted-eval-doc…
eb8680 Jul 28, 2026
4f40407
Add test checking resolution of issue #576
eb8680 Jul 29, 2026
3485df4
harness module
eb8680 Jul 29, 2026
7fea052
Renaming, part 1
eb8680 Jul 29, 2026
e1c26b2
nit
eb8680 Jul 29, 2026
2a8d0c6
nit
eb8680 Jul 29, 2026
9a89558
execution module
eb8680 Jul 29, 2026
6a307d5
persistence module
eb8680 Jul 29, 2026
a2d3272
observability.py
eb8680 Jul 29, 2026
aead502
tracing
eb8680 Jul 29, 2026
58eb532
completions
eb8680 Jul 29, 2026
3eaef0c
retrying.py
eb8680 Jul 29, 2026
190f0d6
choreographies.py
eb8680 Jul 29, 2026
7b22406
TenacityRetryer
eb8680 Jul 29, 2026
9ca087d
rendering.py
eb8680 Jul 29, 2026
524372b
encoding.py
eb8680 Jul 29, 2026
6fe53c8
persisting.py
eb8680 Jul 29, 2026
3130fa7
choreographing.py
eb8680 Jul 29, 2026
f4a5189
scoping.py
eb8680 Jul 29, 2026
b1fb146
coding.py
eb8680 Jul 29, 2026
b96d8ae
providing.py
eb8680 Jul 29, 2026
a1945e9
hooks.py
eb8680 Jul 29, 2026
1a52e83
hooks.py
eb8680 Jul 29, 2026
0bfd814
nit
eb8680 Jul 29, 2026
7260ece
scoping
eb8680 Jul 29, 2026
09232b7
unsafe.py
eb8680 Jul 29, 2026
8b85965
restricted.py
eb8680 Jul 29, 2026
563f442
unsafe -> builtin
eb8680 Jul 29, 2026
6514222
restricted
eb8680 Jul 29, 2026
613a89f
nits
eb8680 Jul 29, 2026
8fd3660
executing
eb8680 Jul 29, 2026
a388657
rename
eb8680 Jul 29, 2026
a0697f6
fix tests
eb8680 Jul 29, 2026
aa16590
Add ClaimCheck llm_example: auditing proof against intent (#660)
eb8680 Jul 29, 2026
6ae402d
Sharpen the ClaimCheck corpus; report that the ablation does not sepa…
eb8680 Jul 29, 2026
4e38c73
Add optimize_anything example: reflective Pareto search over typed ar…
eb8680 Jul 29, 2026
cf0d3cb
__main__
eb8680 Jul 29, 2026
1562eac
Run the ablation on older models at temperature 0; record the result
eb8680 Jul 29, 2026
92c22d6
Fix an answer-key leak into every agent's prompt; withdraw the old re…
eb8680 Jul 29, 2026
bb33b7a
Impoverish the naive arm to match upstream; restore the truncated tax…
eb8680 Jul 29, 2026
26aa158
remove task from investigation.py workspace
eb8680 Jul 29, 2026
dc8ed00
Merge branch 'master' into eb-doctest-synthesis
eb8680 Jul 29, 2026
2aed5de
fix failing doctest
eb8680 Jul 29, 2026
b81172a
remove obsolete notebook
eb8680 Jul 29, 2026
ce245cd
Rebuild the corpus around opaque-invariant projections; the gap still…
eb8680 Jul 29, 2026
01855e0
pack: follow the paper's circle-packing setup instead of approximatin…
eb8680 Jul 29, 2026
6ea0d28
Split the optimize_anything example into a package of three scripts
eb8680 Jul 29, 2026
2e5af20
Rebuild the ClaimCheck corpus as a 1:1 port of upstream's Dafny bench…
eb8680 Jul 29, 2026
ef58e66
Fix port fidelity issues found in review, and isolate the batching me…
eb8680 Jul 29, 2026
c955d07
Correct false claims and measurement defects in prompting.py and kern…
eb8680 Jul 29, 2026
4448adf
Fix two more measurement defects; retract the caveat ablation and its…
eb8680 Jul 29, 2026
2138d99
Strip working history from the example; refresh numbers against shipp…
eb8680 Jul 29, 2026
e469905
Add the missing control, and fix what the numbers were claiming
eb8680 Jul 30, 2026
31b9d77
Record the batching measurement; correct the matched-model comparison
eb8680 Jul 30, 2026
e0cf7c5
Neutralise single-pass module text; attribute the strong baselines to…
eb8680 Jul 30, 2026
31a37de
Withdraw the model-capability explanation for the strong baselines
eb8680 Jul 30, 2026
319a881
Reconcile the two-pass deficit against upstream's same-model runs
eb8680 Jul 30, 2026
b4694a1
Remove a rule leak from the naive arm's prompt; verify all three from…
eb8680 Jul 30, 2026
c2067f8
Correct the claim that all three arms differ by signature
eb8680 Jul 30, 2026
4465ef8
Correct the single-pass attribution and the return-type mechanics
eb8680 Jul 30, 2026
d49acae
Note that upstream measures the single-pass/naive contrast at zero
eb8680 Jul 30, 2026
c82aaca
Drop the single-pass arm; the ablation is two-pass against naive
eb8680 Jul 30, 2026
e726536
Cut the example header to a description of the problem and results
eb8680 Jul 30, 2026
75c9896
Measure the outstanding ablations; none of the three claims reproduce
eb8680 Jul 30, 2026
dd6ab94
optimize_anything example
eb8680 Jul 31, 2026
d057a18
Move the formalization examples into llm_examples/autoformalization
eb8680 Jul 31, 2026
fd52fc1
claimcheck
eb8680 Jul 31, 2026
fe99ee2
rename multi_agent -> multi_agent_choreography
eb8680 Jul 31, 2026
b1779ba
move system prompt builders to scoping
eb8680 Jul 31, 2026
77f1b74
move choreographies to examples
eb8680 Jul 31, 2026
e9ef53c
rename files
eb8680 Jul 31, 2026
de773ab
Merge branch 'master' into eb-doctest-synthesis
eb8680 Jul 31, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 0 additions & 4 deletions .github/workflows/publish_docs.yml
Original file line number Diff line number Diff line change
Expand Up @@ -34,10 +34,6 @@ jobs:
- name: Build docs
run: cd docs && uv run make html

- name: Generate LLM agent reference markdown
run: |
uv run jupyter nbconvert --to markdown docs/source/llm.ipynb --output llm.md --output-dir docs/build/html/

- name: Setup Pages
uses: actions/configure-pages@v4

Expand Down
4 changes: 0 additions & 4 deletions .github/workflows/test_docs.yml
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,3 @@ jobs:
- name: Build docs
run: |
cd docs && uv run make html

- name: Generate LLM agent reference markdown
run: |
uv run jupyter nbconvert --to markdown docs/source/llm.ipynb --output llm.md --output-dir docs/build/html/
7 changes: 0 additions & 7 deletions docs/source/index.rst
Original file line number Diff line number Diff line change
Expand Up @@ -8,13 +8,6 @@ Table of Contents
getting_started
introduction
named_tensor_notation
llm

.. tip::

**For AI coding agents:** A standalone Markdown version of the LLM guide is
available at `llm.md <llm.md>`_ — point your agent there for a complete
reference on building effectful LLM applications.

.. toctree::
:maxdepth: 1
Expand Down
Loading
Loading