Owner

future-agi

1 indexed repository · View on GitHub

future-agi
Open-source, end-to-end platform for evaluating, observing, and improving LLM and AI agent applications. Tracing · Evals · Simulations · ...

Python · 2032 stars

Beginner friendly Everything

Open All

Tool `strict` is stored in the canonical request but has no field in the Anthropic or Gemini tool types Open

bug

future-agi/future-agi#2890 · 1 assignee · Opened Sep 19, 2026
tool_choice "none" keeps the tools, and allowed_tools, parallel_tool_calls and allowed_callers never reach the provider Open

bug

future-agi/future-agi#2889 · 1 assignee · Opened Sep 19, 2026
OpenAI file_id is forwarded as an image URL and a Gemini fileUri, document blocks become empty strings, and file_data and input_audio are passed through untranslated Open

bug

future-agi/future-agi#2888 · 1 assignee · Opened Sep 19, 2026
Falcon AI message rate limiter rejects stop cancellation commands Open

future-agi/future-agi#2878 · 1 assignee · Opened Sep 17, 2026
Falcon AI stream error frames without message_id fail to surface on active message Open

future-agi/future-agi#2877 · 1 assignee · Opened Sep 17, 2026
ai_tools: resolve_experiment() accepts a workspace parameter but never filters by it Open

bug

future-agi/future-agi#2872 · 1 assignee · Opened Sep 17, 2026
docs: deploy verify step uses wrong backend health endpoint (/healthz 404s, should be /health/) Open

bug

future-agi/future-agi#2869 · 1 assignee · Opened Sep 17, 2026
docs: INSTALLATION lists wrong MinIO host ports (9000/9001 instead of 9005/9006) Open

bug

future-agi/future-agi#2867 · 1 assignee · Opened Sep 17, 2026
docs: add Chinese README translation Open

future-agi/future-agi#2852 · 1 assignee · Opened Sep 16, 2026
tracer: unconditional import fcntl in mutation_lock.py breaks every Windows Django test run Open

bug

future-agi/future-agi#2844 · 1 comment · 1 assignee · Opened Sep 16, 2026
All four aggregate projections on spans store a doubled -State, so no naturally-written aggregate can match them and every such query silently full-scans Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

future-agi/future-agi#2842 · Opened Sep 16, 2026
Trace-view graphs count spans while the trace list counts traces: on COL the graph renders ~119x the traces beneath it Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

future-agi/future-agi#2841 · Opened Sep 16, 2026
ai_tools: resolve_experiment() filters ExperimentsTable by a non-existent 'organization' field Open

bug

future-agi/future-agi#2840 · 1 assignee · Opened Sep 16, 2026
spans_hourly_rollup counts insert deliveries, not retained spans: unfiltered Observe graphs and dashboard widgets render up to 3.5x the base table, undisclosed Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

future-agi/future-agi#2839 · 4 comments · Opened Sep 16, 2026
[Good First Issue] Prompt editor: 'Select All' (Ctrl+A) selects embedded images/audio/PDFs along with text Open

Difficulty 3/5 1-2 days Newbie friendliness 74/100

future-agi/future-agi#2836 · 3 comments · Opened Sep 16, 2026
Seven test modules default to a forwarded ClickHouse port and write to whatever answers Open

Difficulty 3/5 1-2 days Newbie friendliness 25/100

future-agi/future-agi#2830 · Opened Sep 15, 2026
test_ch25_typed_json_roundtrip has never run in CI, and its column-type contract may conflict with the test compose file Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

future-agi/future-agi#2829 · 1 comment · Opened Sep 15, 2026
Substring filters in non-Latin scripts get no index acceleration (ASCII-lowered ngram anchor declines them) Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

future-agi/future-agi#2825 · Opened Sep 15, 2026
feat: add CI triage for grouping related issues and pull requests Open

future-agi/future-agi#2816 · 1 assignee · Opened Sep 15, 2026
Observe: Session list replays latest state across the whole request window, so a 30-day page never returns Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

future-agi/future-agi#2810 · Opened Sep 15, 2026
Observe: filter values are inlined several times per statement with no ceiling on value size Open

Difficulty 5/5 Over a week Newbie friendliness 32/100

future-agi/future-agi#2802 · Opened Sep 15, 2026
Notification sanitizer uses regex blacklist on dangerouslySetInnerHTML sink Open

future-agi/future-agi#2795 · 1 assignee · Opened Sep 15, 2026
Auto annotation fails on every run as Cell has no col_id field. Open

bug

future-agi/future-agi#2792 · 1 comment · 1 assignee · Opened Sep 14, 2026
simulate_chat loops forever when models never say "topic resolved" Open

bug

future-agi/future-agi#2790 · 1 comment · 1 assignee · Opened Sep 14, 2026
Self-host images ship ~6.9 GB of CUDA libraries the compose stack never uses Open

accepted bug

future-agi/future-agi#2776 · 2 comments · 1 assignee · Opened Sep 13, 2026
Gateway: Add Provider dialog - Multiselect adds unclicked models on blur Open

bug

future-agi/future-agi#2760 · 1 comment · 1 assignee · Opened Sep 12, 2026
agentic_eval: calculate_rouge (RougeScore) returns result as a string, so it can never be flagged as a failure Open

future-agi/future-agi#2738 · 1 assignee · Opened Sep 11, 2026
Observe: attribute-filtered trace list never loads on high-volume projects (short-string SPAN_ATTRIBUTE filters) Open

Difficulty 5/5 Over a week Newbie friendliness 10/100

future-agi/future-agi#2734 · 10 comments · Opened Sep 11, 2026
Environment: hosted, sandboxed end-to-end simulation for repository-backed and provider-hosted agents Open

accepted enhancement feature

future-agi/future-agi#2725 · 1 assignee · Opened Sep 11, 2026
agentcc-gateway README: the "fully reproducible" benchmark harness (bench/) is not in the repo Open

future-agi/future-agi#2705 · 1 comment · 1 assignee · Opened Sep 10, 2026
Self-host install fails on Apple Silicon: backend image is amd64-only, and property-catalog-supervisor's 768M cap kills it during first boot Open

future-agi/future-agi#2704 · 1 assignee · Opened Sep 10, 2026
Add a tested audio-metrics extractor for pitch and estimated SNR Open

accepted backend enhancement good first issue simulate

Difficulty 3/5 1-2 days Newbie friendliness 76/100

future-agi/future-agi#2696 · 2 comments · Opened Sep 10, 2026
Circuit breaker half-open state does not limit probe traffic Open

bug

future-agi/future-agi#2693 · 1 comment · 1 assignee · Opened Sep 10, 2026
A node that fails after emitting partial output shows no error in the run panel Open

future-agi/future-agi#2685 · 1 assignee · Opened Sep 10, 2026
Dataset and experiment row runs record only a final value, with no execution trace Open

backend datasets enhancement good first issue

Difficulty 4/5 3-5 days Newbie friendliness 55/100

future-agi/future-agi#2665 · 2 comments · Opened Sep 9, 2026
A graph created from a trace keeps no link back to the spans it came from Open

agent-playground backend enhancement good first issue

Difficulty 4/5 3-5 days Newbie friendliness 58/100

future-agi/future-agi#2664 · 2 comments · Opened Sep 9, 2026
Agent run panel ships a node execution list that is never rendered Open

agent-playground bug frontend good first issue

Difficulty 3/5 1-2 days Newbie friendliness 76/100

future-agi/future-agi#2663 · 1 comment · Opened Sep 9, 2026
Simulate rejects Twilio call data even though Twilio is a supported provider Open

backend bug good first issue simulate

Difficulty 4/5 3-5 days Newbie friendliness 55/100

future-agi/future-agi#2662 · 1 comment · Opened Sep 9, 2026
Experiments render every prompt as Jinja, ignoring the format the prompt was saved with Open

backend bug frontend good first issue

Difficulty 4/5 3-5 days Newbie friendliness 72/100

future-agi/future-agi#2661 · 1 comment · Opened Sep 9, 2026
Self-hosted stack fails to start on Windows: shell scripts checked out with CRLF Open

bug

future-agi/future-agi#2656 · 1 assignee · Opened Sep 9, 2026
Carry billing cadence/display unit in the pricing API Open

future-agi/future-agi#2653 · 1 assignee · Opened Sep 9, 2026
chore: back-merge `main` into `dev` after the 1.37.0 release Open

accepted

Difficulty 2/5 1-3 hours Newbie friendliness 68/100

future-agi/future-agi#2649 · Opened Sep 9, 2026
Percentage Change alert thresholds scale the standard deviation, not the previous period, and collapse onto the mean when stddev is 0 Open

future-agi/future-agi#2636 · 2 comments · 1 assignee · Opened Sep 8, 2026
Custom-model pricing cache is never invalidated: new or edited pricing is ignored for up to 24 hours Open

future-agi/future-agi#2635 · 2 comments · 1 assignee · Opened Sep 8, 2026
Custom-model costs are 1000x too high: UI collects price per million tokens, backend divides by 1000 Open

future-agi/future-agi#2634 · 1 assignee · Opened Sep 8, 2026
Compare drawer cost/latency chips misspell height as heigth, so the style is silently dropped Open

future-agi/future-agi#2633 · 2 comments · 1 assignee · Opened Sep 8, 2026
Alert-monitor log timestamps use the server's local clock labelled as UTC, and resolved_at is stored naive Open

future-agi/future-agi#2632 · 1 comment · 1 assignee · Opened Sep 8, 2026
Support custom API path prefixes for UI-managed gateway providers Open

accepted

Difficulty 4/5 3-5 days Newbie friendliness 65/100

future-agi/future-agi#2628 · 1 comment · Opened Sep 8, 2026
Cap memory per service in the self-host stack so one container can't claim the whole machine Open

feature

future-agi/future-agi#2616 · 1 assignee · Opened Sep 8, 2026
Perf: FID computation every frame and a potential re-invocation overhead Open

feature

future-agi/future-agi#2598 · 1 comment · 1 assignee · Opened Sep 7, 2026

Showing the newest 100

This page lists what was indexed most recently. Advanced filter has the whole inventory, narrowed by language, difficulty and how long a task takes.

Open advanced filter

future-agi

Showing the newest 100

Get new issues in your inbox