[AI] Centralize default max_tokens by sjsjsjjs534 · Pull Request #205 · bitflicker64/Termstory

sjsjsjjs534 · 2026-07-02T13:39:17Z

Summary

This PR centralizes the hardcoded 1500 token limit used across five long-form AI generator functions into a single exported DEFAULT_MAX_TOKENS constant in termstory/ai.py, improving maintainability and ensuring consistent default behavior across all generators.

Changes

Core

termstory/ai.py: Added module-level constant DEFAULT_MAX_TOKENS: int = 1500 to centralize the default token limit for long-form AI generators.
termstory/ai.py: Updated _send_llm_request function's default parameter from max_tokens: int = 1500 to max_tokens: int = DEFAULT_MAX_TOKENS.
termstory/ai.py: Updated generate_timeframe_summary to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated generate_wrapped_summary to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated translate_git_anger to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated predict_bugs_from_sessions to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated generate_rpg_bio to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.

Tests

tests/test_ai.py: Added imports for sys, SimpleNamespace, and import termstory.ai as ai to support the new regression test.
tests/test_ai.py: Updated existing test test_timeframe_summary_max_tokens to reference ai.DEFAULT_MAX_TOKENS instead of the hardcoded value 1500 in the assertion.
tests/test_ai.py: Added new regression test test_long_form_generators_share_default_max_tokens that patches DEFAULT_MAX_TOKENS to 1234 and verifies all five long-form generators (generate_timeframe_summary, generate_wrapped_summary, translate_git_anger, predict_bugs_from_sessions, generate_rpg_bio) pick up the changed value.

Fixes

Fixes [Easy] ai.py: Hardcoded max_tokens=1500 duplicated across 5 call sites #192 — Centralizes hardcoded token limit into a shared constant for better maintainability

Breaking Changes

None

Testing

Run the AI test suite with isolated config directories:

XDG_CONFIG_HOME=/private/tmp/termstory-test-config XDG_DATA_HOME=/private/tmp/termstory-test-data .venv/bin/python -m pytest tests/test_ai.py -q

Run the full test suite:

XDG_CONFIG_HOME=/private/tmp/termstory-test-config XDG_DATA_HOME=/private/tmp/termstory-test-data .venv/bin/python -m pytest tests/ -q

Notes

The constant DEFAULT_MAX_TOKENS is exported at the module level, making it available for external configuration or testing purposes. The regression test uses monkeypatch.setattr with raising=False to override the constant at runtime, ensuring that all five generators consistently reference the shared default rather than hardcoded values.

git-miku · 2026-07-02T13:39:24Z

👋 Thanks for the PR, @sjsjsjjs534!

To get this PR reviewed, please ⭐ star this repo — it's free and helps others discover it. Once starred, run :miku /check-star on this PR and I'll start helping out.

A do-not-merge/needs-star label has been applied to block merge until then.

greptile-apps · 2026-07-02T13:42:03Z

Greptile Summary

This PR replaces five hardcoded 1500 token limits across long-form AI generators with a single exported DEFAULT_MAX_TOKENS = 1500 constant in termstory/ai.py, and updates the test suite to verify the shared constant is picked up dynamically.

termstory/ai.py: Adds DEFAULT_MAX_TOKENS: int = 1500 at module scope; updates _send_llm_request's default parameter and the five call sites in generate_timeframe_summary, generate_wrapped_summary, translate_git_anger, predict_bugs_from_sessions, and generate_rpg_bio.
tests/test_ai.py: Converts the existing test_timeframe_summary_max_tokens assertion to reference ai.DEFAULT_MAX_TOKENS and adds test_long_form_generators_share_default_max_tokens, which monkeypatches the constant to 1234 and verifies all five generators propagate the override end-to-end.

Confidence Score: 5/5

Safe to merge — purely mechanical replacement of five hardcoded literals with a named constant; no logic changes.

Every target call site was updated and verified against the diff. The two intentionally distinct callers (500 and 2000 tokens) are correctly left unchanged. The new regression test exercises dynamic module-level lookup for all five generators, confirming the constant is not baked in at compile time. No behaviour changes in production paths.

No files require special attention.

Important Files Changed

Filename	Overview
termstory/ai.py	Adds DEFAULT_MAX_TOKENS constant and updates all five target call sites; no remaining hardcoded 1500 in generator bodies; two intentionally distinct call sites (500 and 2000) are correctly left untouched.
tests/test_ai.py	Existing max_tokens assertion updated to reference the constant; new regression test correctly verifies all five generators do a module-level lookup (not a compile-time bind) of DEFAULT_MAX_TOKENS.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    C[DEFAULT_MAX_TOKENS = 1500]
    C --> A[_send_llm_request\ndefault param]
    C --> B1[generate_timeframe_summary]
    C --> B2[generate_wrapped_summary]
    C --> B3[translate_git_anger]
    C --> B4[predict_bugs_from_sessions]
    C --> B5[generate_rpg_bio]

    B1 -->|max_tokens=DEFAULT_MAX_TOKENS| R[_send_llm_request]
    B2 -->|max_tokens=DEFAULT_MAX_TOKENS| R
    B3 -->|max_tokens=DEFAULT_MAX_TOKENS| R
    B4 -->|max_tokens=DEFAULT_MAX_TOKENS| R
    B5 -->|max_tokens=DEFAULT_MAX_TOKENS| R

    X1[generate_ai_summary] -->|max_tokens=500| R
    X2[generate_daily_chronicle] -->|max_tokens=2000| R

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    C[DEFAULT_MAX_TOKENS = 1500]
    C --> A[_send_llm_request\ndefault param]
    C --> B1[generate_timeframe_summary]
    C --> B2[generate_wrapped_summary]
    C --> B3[translate_git_anger]
    C --> B4[predict_bugs_from_sessions]
    C --> B5[generate_rpg_bio]

    B1 -->|max_tokens=DEFAULT_MAX_TOKENS| R[_send_llm_request]
    B2 -->|max_tokens=DEFAULT_MAX_TOKENS| R
    B3 -->|max_tokens=DEFAULT_MAX_TOKENS| R
    B4 -->|max_tokens=DEFAULT_MAX_TOKENS| R
    B5 -->|max_tokens=DEFAULT_MAX_TOKENS| R

    X1[generate_ai_summary] -->|max_tokens=500| R
    X2[generate_daily_chronicle] -->|max_tokens=2000| R

_{Reviews (3): Last reviewed commit: "Potential fix for pull request finding" | Re-trigger Greptile}

greptile-apps · 2026-07-02T13:42:07Z

+    monkeypatch.setattr(urllib.request, "urlopen", mock_urlopen)
+    monkeypatch.setattr("termstory.config.load_config", lambda: {})
+    monkeypatch.setattr(ai, "_get_all_active_project_contexts", lambda: [])
+    monkeypatch.setattr(ai, "DEFAULT_MAX_TOKENS", 1234, raising=False)


The raising=False flag is not needed here since DEFAULT_MAX_TOKENS is definitively exported from termstory.ai. With raising=False, a typo in the attribute name (e.g. DEFAULT_MAX_TOKEN) would silently set a new attribute on the module instead of patching the existing one, the generators would still read the old 1500 value, and all five assertions would fail with 1500 != 1234 — but the monkeypatch itself would give no warning. Dropping the flag makes the intent clear and lets pytest raise immediately if the name is ever wrong.

Suggested change

monkeypatch.setattr(ai, "DEFAULT_MAX_TOKENS", 1234, raising=False)

monkeypatch.setattr(ai, "DEFAULT_MAX_TOKENS", 1234)

bitflicker64 · 2026-07-02T13:57:57Z

miku /check-star

git-miku · 2026-07-02T13:58:08Z

👋 Thanks for the PR, @sjsjsjjs534!

To get this PR reviewed, please ⭐ star this repo — it's free and helps others discover it. Once starred, run :miku /check-star on this PR and I'll start helping out.

A do-not-merge/needs-star label has been applied to block merge until then.

bitflicker64 · 2026-07-02T13:58:52Z

/rewrite

git-miku · 2026-07-02T13:59:27Z

miku /rewrite

✅ PR description rewritten! Check the PR body above.

Preview

Summary

This PR centralizes the hardcoded 1500 token limit used across five long-form AI generator functions into a single exported DEFAULT_MAX_TOKENS constant in termstory/ai.py, improving maintainability and ensuring consistent default behavior across all generators.

Changes

Core

termstory/ai.py: Added module-level constant DEFAULT_MAX_TOKENS: int = 1500 to centralize the default token limit for long-form AI generators.
termstory/ai.py: Updated _send_llm_request function's default parameter from max_tokens: int = 1500 to max_tokens: int = DEFAULT_MAX_TOKENS.
termstory/ai.py: Updated generate_timeframe_summary to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated generate_wrapped_summary to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated translate_git_anger to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated predict_bugs_from_sessions to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.
termstory/ai.py: Updated generate_rpg_bio to use DEFAULT_MAX_TOKENS instead of hardcoded 1500 when calling _send_llm_request.

Tests

tests/test_ai.py: Added imports for sys, SimpleNamespace, and import termstory.ai as ai to support the new regression test.
tests/test_ai.py: Updated existing test test_timeframe_summary_max_tokens to reference ai.DEFAULT_MAX_TOKENS instead of the hardcoded value 1500 in the assertion.
tests/test_ai.py: Added new regression test test_long_form_generators_share_default_max_tokens that patches DEFAULT_MAX_TOKENS to 1234 and verifies all five long-form generators (generate_timeframe_summary, generate_wrapped_summary, translate_git_anger, predict_bugs_from_sessions, generate_rpg_bio) pick up the changed value.

Fixes

Fixes [Easy] ai.py: Hardcoded max_tokens=1500 duplicated across 5 call sites #192 — Centralizes hardcoded token limit into a shared constant for better maintainability

Breaking Changes

None

Testing

Run the AI test suite with isolated config directories:

XDG_CONFIG_HOME=/private/tmp/termstory-test-config XDG_DATA_HOME=/private/tmp/termstory-test-data .venv/bin/python -m pytest tests/test_ai.py -q

Run the full test suite:

XDG_CONFIG_HOME=/private/tmp/termstory-test-config XDG_DATA_HOME=/private/tmp/termstory-test-data .venv/bin/python -m pytest tests/ -q

Notes

The constant DEFAULT_MAX_TOKENS is exported at the module level, making it available for external configuration or testing purposes. The regression test uses monkeypatch.setattr with raising=False to override the constant at runtime, ensuring that all five generators consistently reference the shared default rather than hardcoded values.

bitflicker64 · 2026-07-03T17:41:45Z

/triage

git-miku · 2026-07-03T17:41:52Z

miku /triage

✅ Marked as triaged (triage/triaged applied, needs-triage removed).

bitflicker64 · 2026-07-03T17:45:56Z


 # Module-level defaults — used when config is unavailable.
+DEFAULT_MAX_TOKENS: int = 1500
 _DEFAULT_MAX_FAILURES: int = 3


Suggested change

_DEFAULT_MAX_FAILURES: int = 3

# Module-level defaults — used when config is unavailable.

# Note: DEFAULT_MAX_TOKENS is public (no underscore prefix) because tests

# override it via monkeypatch.setattr. The other defaults stay private.

DEFAULT_MAX_TOKENS: int = 1500

_DEFAULT_MAX_FAILURES: int = 3

_DEFAULT_COOLDOWN_SECONDS: float = 60.0

The naming is inconsistent: DEFAULT_MAX_TOKENS (public) vs _DEFAULT_MAX_FAILURES (private). Either the public one should also be private, or the others should be public. Currently DEFAULT_MAX_TOKENS is the only public constant and the convention isn't clear.

bitflicker64 · 2026-07-03T17:45:56Z

+        resp_payload = {"choices": [{"message": {"content": "ok"}}]}
+        return MockResponse(json.dumps(resp_payload).encode("utf-8"))
+
+    monkeypatch.setattr(urllib.request, "urlopen", mock_urlopen)


The new test test_long_form_generators_share_default_max_tokens (75 lines) overlaps with the existing test_default_max_tokens_propagates_to_request_body test, which already verifies that DEFAULT_MAX_TOKENS propagates to the request body. The new one exercises generate_rpg_bio and 4 other generators — that's useful but could be folded into the existing test as additional assertions rather than a separate test that mostly duplicates the assertion pattern.

Copilot

Pull request overview

This PR centralizes the long-form AI generators’ default max_tokens value by introducing a single exported constant (DEFAULT_MAX_TOKENS) in termstory/ai.py, then updates the generators and tests to reference it for consistency and easier tuning.

Changes:

Added DEFAULT_MAX_TOKENS in termstory/ai.py and replaced hardcoded 1500 at the five long-form generator call sites.
Updated _send_llm_request’s default max_tokens parameter to reference the new constant.
Updated tests/test_ai.py to assert against ai.DEFAULT_MAX_TOKENS and added a regression test intended to ensure all long-form generators share the same default.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
termstory/ai.py	Introduces `DEFAULT_MAX_TOKENS` and updates long-form generator calls (and `_send_llm_request` default) to use it.
tests/test_ai.py	Updates assertions to use the constant and adds a regression test for shared defaults.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

    model_name: str,
    provider: str,
-    max_tokens: int = 1500,
+    max_tokens: int = DEFAULT_MAX_TOKENS,
    timeout: float = 30.0
 ) -> Optional[str]:


git-miku · 2026-07-06T22:24:35Z

👋 Thanks for the PR, @sjsjsjjs534!

To get this PR reviewed, please ⭐ star this repo — it's free and helps others discover it. Once starred, run :miku /check-star on this PR and I'll start helping out.

A do-not-merge/needs-star label has been applied to block merge until then.

bitflicker64 · 2026-07-06T22:25:54Z

@copilot resolve the merge conflicts in this pull request

greptile-apps · 2026-07-07T16:15:38Z

No reviewable files after applying ignore patterns.

git-miku · 2026-07-07T16:15:40Z

👋 Thanks for the PR, @sjsjsjjs534!

To get this PR reviewed, please ⭐ star this repo — it's free and helps others discover it. Once starred, run :miku /check-star on this PR and I'll start helping out.

A do-not-merge/needs-star label has been applied to block merge until then.

git-miku Bot added the do-not-merge/needs-star PR author has not starred this repo — run /check-star after starring label Jul 2, 2026

greptile-apps Bot reviewed Jul 2, 2026

View reviewed changes

bitflicker64 closed this Jul 2, 2026

bitflicker64 reopened this Jul 2, 2026

git-miku Bot added the triage/triaged This PR has been triaged by a maintainer label Jul 3, 2026

bitflicker64 reviewed Jul 3, 2026

View reviewed changes

Comment thread termstory/ai.py

bitflicker64 reviewed Jul 3, 2026

View reviewed changes

bitflicker64 requested a review from Copilot July 6, 2026 22:17

Copilot started reviewing on behalf of bitflicker64 July 6, 2026 22:17 View session

Copilot AI reviewed Jul 6, 2026

View reviewed changes

Comment thread termstory/ai.py

Comment on lines 168 to 172

model_name: str,

provider: str,

max_tokens: int = 1500,

max_tokens: int = DEFAULT_MAX_TOKENS,

timeout: float = 30.0

) -> Optional[str]:

Comment thread tests/test_ai.py Outdated

Copilot AI mentioned this pull request Jul 6, 2026

Resolve _send_llm_request default max_tokens binding to support runtime overrides #235

Merged

sjsjsjjs534 closed this Jul 7, 2026

sjsjsjjs534 force-pushed the codex/issue-192-default-max-tokens branch from 948eed3 to 58363e0 Compare July 7, 2026 16:15

	monkeypatch.setattr(ai, "DEFAULT_MAX_TOKENS", 1234, raising=False)
	monkeypatch.setattr(ai, "DEFAULT_MAX_TOKENS", 1234)

Conversation

sjsjsjjs534 commented Jul 2, 2026 • edited by git-miku Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Core

Tests

Fixes

Breaking Changes

Testing

Notes

Uh oh!

git-miku Bot commented Jul 2, 2026

Uh oh!

greptile-apps Bot commented Jul 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 5/5

Important Files Changed

Flowchart

Uh oh!

greptile-apps Bot Jul 2, 2026

Choose a reason for hiding this comment

Uh oh!

bitflicker64 commented Jul 2, 2026

Uh oh!

git-miku Bot commented Jul 2, 2026

Uh oh!

bitflicker64 commented Jul 2, 2026

Uh oh!

git-miku Bot commented Jul 2, 2026

Summary

Changes

Core

Tests

Fixes

Breaking Changes

Testing

Notes

Uh oh!

bitflicker64 commented Jul 3, 2026

Uh oh!

git-miku Bot commented Jul 3, 2026

Uh oh!

Uh oh!

bitflicker64 Jul 3, 2026

Choose a reason for hiding this comment

Uh oh!

bitflicker64 Jul 3, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

git-miku Bot commented Jul 6, 2026

Uh oh!

bitflicker64 commented Jul 6, 2026

Uh oh!

greptile-apps Bot commented Jul 7, 2026

Uh oh!

git-miku Bot commented Jul 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

sjsjsjjs534 commented Jul 2, 2026 •

edited by git-miku Bot

Loading

greptile-apps Bot commented Jul 2, 2026 •

edited

Loading