## User Feedback Analysis — 2026-05-05 (based on 2026-05-02 to 2026-05-04 community + GitHub signals)

### Data notes (scope & sample size)
- Daily community summaries available for **Discord (2026-05-02/03/04)** and a **GitHub month snapshot (May 2026)** including several high-detail issues.
- This report aggregates the explicitly stated friction points across those sources (≈**15 distinct “feedback atoms”**: Discord questions/complaints + GitHub issues + “identified issues” list in the daily dev summary).

---

## 1) Pain Point Categorization (Top recurring 5–7)

### 1) **Integration — Telegram connector reliability + routing**
**Severity:** High (silent message loss / missing dispatch)  
**Frequency in sample:** ~**27%** (4/15 atoms)  
**Evidence (examples):**
- “**Telegram wrapper bypassing agent runtime and preventing skill and action dispatch**” (daily dev “Identified Issues”, 2026-05-04).
- GitHub issue: **dual pollers** (`@elizaos/plugin-telegram` + Milady wrapper) cause **silent message loss** (every other message dropped) (Sw4pIO, #7245).
- GitHub issue: Telegram wrapper bypasses runtime (also reflected in daily summary).

**Who it affects most:** builders shipping Telegram bots (Milady and other consumer apps), newcomers doing “first connector” setup.

---

### 2) **Technical Functionality — Auth/subscription detection and “it should work but doesn’t”**
**Severity:** High (401s, “install required” false negatives; blocks onboarding)  
**Frequency:** ~**20%** (3/15)  
**Evidence:**
- Anthropic OAuth subscription path broken due to missing preload file (`claude-code-stealth.mjs`) on fresh clone; results in **401 Invalid authentication credentials** (Sw4pIO, #7210).
- Codex CLI auth detection misses modern `tokens.access_token` shape → UI reports **“install required”** incorrectly (Sw4pIO, #7243).
- Cloud auth/session issues were actively being stabilized (multiple Cloud login/session sync fixes referenced in 2026-05-04 dev summary).

**Who it affects most:** self-hosters and contributors cloning repos fresh; anyone using subscription/OAuth paths vs API keys.

---

### 3) **Technical Functionality — Database/schema drift in plugin-sql migrations**
**Severity:** High (chat becomes unusable on fresh DB; cascading errors)  
**Frequency:** ~**7%** (1/15) but high impact  
**Evidence:**
- Missing drizzle `pgTable` defs → tables never created → repeated query failures → state composition incomplete → user-visible parsing errors (Sw4pIO, #7222).

**Who it affects most:** new installs using PGLite / plugin-sql; anyone evaluating “works out of the box.”

---

### 4) **Performance/UX — UI instability under load (flicker) + high-volume event handling**
**Severity:** Medium–High (breaks trust, “feels broken”)  
**Frequency:** ~**7%** (1/15)  
**Evidence:**
- “**Activity panel flickering under high-volume WebSocket events**” (daily dev “Identified Issues”, 2026-05-04).

**Who it affects most:** power users running many agents/events; LifeOps users with continuous streams.

---

### 5) **Technical Functionality — Permissions/confirmation flow not respecting bypass flags**
**Severity:** Medium–High (blocks automation; confusing safety model)  
**Frequency:** ~**7%** (1/15)  
**Evidence:**
- “**dangerously-skip-permissions flag failing to bypass confirmation loops**” (daily dev “Identified Issues”, 2026-05-04).

**Who it affects most:** developers trying to run headless/automated flows; CI-style usage.

---

### 6) **Community / UX — Discord moderation friction (bots) + spam filter blocking normal URLs**
**Severity:** Medium (interaction tax; reduces support quality)  
**Frequency:** ~**20%** (3/15)  
**Evidence:**
- Complaint: “**Nigerian bots getting out of hand in the chat**”; mod acknowledges active management (Discord, 2026-05-04).
- URL sharing blocked by strict filter; workaround is wrapping links in backticks (Discord, 2026-05-03 help interaction: sentient_dawn ↔ odilitime).
- General noise from price-talk spikes makes technical questions easier to miss (implicit in channel focus).

**Who it affects most:** newcomers asking setup questions; helpers trying to provide links; moderators.

---

### 7) **Communication Gap — Token/price expectations vs “builders, not investors” positioning**
**Severity:** Medium (expectation mismatch fuels churn/hostility)  
**Frequency:** ~**13%** (2/15)  
**Evidence:**
- Repeated price-focus: V3 up **90–130%**, comparisons to other tokens (e.g., Zerebro +600%), market cap references (Discord, 2026-05-04).
- Shaw explicitly clarifies: holders aren’t “traditional investors”; buybacks require revenue; long business timelines (Discord, 2026-05-03).
- User confusion: “What happened to cause the price movement?” and “confused about low liquidity on Solana” (Discord questions, 2026-05-04).

**Who it affects most:** speculator-heavy segments entering via token, not product.

---

## 2) Usage Pattern Analysis (actual vs intended)

### Observed “actual usage”
1. **Trading agents as a primary entry use-case**
   - Example: marianodim building a **spot trading agent** for SOL/SUI + 5 tokens, aiming for 1–2 signals/week (Discord, 2026-05-02).
   - Social copy-trading direction: `plugin-social-alpha` analyzes “trenches” channels for profitability (Shaw, 2026-05-03).

2. **Connector-first adoption**
   - Users appear to start with Telegram/Discord/Slack connectors and only then explore deeper agent capabilities—making connector reliability a “first impression” dependency.

3. **Cloud + monetization as an emerging path**
   - Strong engineering push toward **monetized container app domains**, Cloudflare management, app-scoped chat, and billing stability (dev summary + Cloud PRs on 2026-05-04).

### Gaps vs intended usage
- Intended: “open-source agent framework for builders.”  
  Actual: a significant portion of Discord discussion is **token performance** and liquidity logistics, which can crowd out onboarding/product adoption.

### Feature requests that align with real usage patterns
- **“One-click reliable Telegram bot”** mode (mutual exclusion of pollers; clear ownership of token/connector runtime).
- **Trading-agent templates** (spot trading starter, social-alpha/copy-trade patterns) since community builders are already shipping these.

---

## 3) Implementation Opportunities (2–3 per major pain point, prioritized)

Below: **(Impact / Difficulty)** with concrete actions.

### A) Telegram integration reliability (silent loss + bypassed runtime)
1) **Enforce single-consumer ownership per bot token** (High / Low–Med)  
   - Add a startup guard: if wrapper polling is enabled, **auto-disable** plugin polling (or vice versa) and log loudly.  
   - Similar pattern: Home Assistant integrations enforce “one coordinator per device/account” and warn on duplicates.

2) **Route all Telegram updates through agent runtime consistently** (High / Med)  
   - Remove/retire wrapper once upstream plugin is fixed; or refactor wrapper to call the same dispatch path as the plugin.  
   - Similar approach: Bot frameworks (e.g., Hubot adapters) normalize inbound events into a single message bus.

3) **Add an integration test: “send 10 messages, receive 10 replies”** (High / Low)  
   - Deterministic harness using Telegram sandbox/mocks; fail builds on drops.  
   - Similar approach: Matrix/Slack adapters in many OSS bots rely on replayable fixtures for delivery guarantees.

---

### B) Auth/subscription detection (Anthropic/Codex) + missing files on fresh clone
1) **Fail loud on missing auth preloads / required artifacts** (High / Low)  
   - When `stealth.claude` is enabled but preload file missing, show a red banner + actionable steps.  
   - Similar approach: Next.js and Vite commonly hard-error on missing critical build artifacts rather than silently skipping.

2) **Support multiple credential shapes with migration helpers** (High / Low)  
   - Update Codex auth detection to recognize modern `tokens.access_token` format (#7243).  
   - Add an importer that writes into `~/.eliza/auth/*` consistently.

3) **Publish “Auth Matrix” doc + troubleshooting table** (Med / Low)  
   - A single page listing: API key vs OAuth subscription per provider, required env vars, expected files, common 401 causes.

---

### C) plugin-sql schema drift / missing tables on fresh DB
1) **Schema source-of-truth validation at boot** (High / Med)  
   - On plugin-sql init, compare abstract schema exports vs drizzle tables; **error** if mismatch (as suggested in #7222).  
   - Similar approach: Prisma validates schema migrations; Django checks migration graph consistency.

2) **Add “fresh DB smoke test” in CI** (High / Low)  
   - Boot with empty PGLite, run one chat turn, assert no missing-table queries.

3) **Generate drizzle tables from abstract schemas (or deprecate one)** (Med / High)  
   - Long-term: remove dual-schema drift vector.

---

### D) Cloud monetized chat / billing correctness + auth error codes
1) **Normalize auth errors to 401/403 everywhere** (High / Low)  
   - Fix Promise.all catch-path that converts auth errors into 500s (noted in Greptile review on Cloud PR #7376).  
   - Similar approach: Stripe-like APIs strictly preserve error classes to prevent client retry confusion.

2) **Make credit reconciliation failure-safe** (High / Med)  
   - Ensure: if response content delivered, do not refund to zero; if reconciliation fails after provider response, return response and schedule reconciliation retry.  
   - Similar approach: idempotent billing with a job queue (e.g., Temporal/Celery) for post-response settlement.

3) **Add a billing “ledger invariants” test suite** (Med / Med)  
   - Property tests: “credits never negative,” “delivered tokens must be charged eventually,” “no free inference on DB blip.”

---

### E) Activity panel flicker under websocket load
1) **Debounce/coalesce activity events in UI** (Med / Low)  
   - Use requestAnimationFrame batching or a fixed-rate reducer.  
   - Similar approach: log viewers (Grafana/Loki UIs) batch updates to reduce repaint storms.

2) **Backpressure on websocket stream** (Med / Med)  
   - Drop non-critical events when saturated; provide “paused / live” toggle.

---

### F) Discord moderation + spam filter usability
1) **Replace blanket URL blocking with staged trust levels** (High / Low)  
   - Allow URLs from trusted roles (Verified/Contributor) and allowlist domains (github.com, docs, elizaos.*).  
   - Similar approach: many OSS Discords use AutoMod + role-based exemptions.

2) **Add a “Support Intake” channel or forum-mode posts** (Med / Low)  
   - Separate token discussion from technical support; reduce missed help requests.

3) **Bot reporting workflow** (Med / Low)  
   - Single command / button to flag suspected bots; auto-hide until mod review.

---

### G) Token/roadmap expectation mismatch
1) **Pin a “What elizaOS is / isn’t” message + quarterly milestones** (Med / Low)  
   - Reinforce: revenue-first, buybacks only after revenue, expected timelines.

2) **Create a “Market talk containment” structure** (Med / Low)  
   - Dedicated channel(s) for price/liquidity so technical channels remain usable during volatility spikes.

---

## 4) Communication Gaps (expectations vs reality)

### Recurring mismatches
- **“Token holders = investors” vs team stance**  
  - Shaw clarifies holders are not traditional investors; buybacks require revenue (2026-05-03).  
  - Users still ask for catalysts and compare to other token pumps (2026-05-04).

- **“Connectors should be plug-and-play” vs current integration complexity**  
  - Telegram reliability issues (dual pollers, bypassed runtime) violate user expectations because failures are **silent**.

### Recurring questions indicating onboarding/doc gaps
- “**Why is Solana liquidity low?**” and “**What caused price movement?**” (Discord, 2026-05-04)  
  Suggests a missing explainer on:
  - where canonical liquidity is,
  - official contract addresses / chains,
  - bridges/wrapped assets (if applicable),
  - risks of thin liquidity venues.

### Suggested expectation-alignment improvements
- Add “**Known Issues & Status**” page (or a pinned Discord message) covering:
  - Telegram current status + recommended config,
  - auth subscription limitations (Anthropic/Codex),
  - cloud login/session known gotchas.
- Publish “**Getting Started: First Successful Agent**” path that avoids flaky surfaces until stabilized (e.g., start with local chat + one connector, not two pollers).

---

## 5) Community Engagement Insights

### Power users observed (and their needs)
- **Sw4pIO (GitHub issues):** extremely high-signal bug reports with repro steps and proposed fixes.  
  **Need:** faster triage loop + “verified reporter” label + invitation to help maintain integration test suites.
- **odilitime (Discord helper/mod + docs/roadmap work):** handles support + ops (spam filter, roadmap updates).  
  **Need:** better tooling to reduce manual moderation/support burden (role-based URL policy, support channel separation).
- **NubsCarson / standujar / 2-A-M (core contributors):** shipping Cloud, auth, and automation UX.  
  **Need:** clearer “definition of done” gates for high-risk merges (billing invariants, auth status codes).

### Newcomer questions (onboarding friction signals)
- Liquidity/where-to-buy confusion and price-catalyst questions suggest many newcomers arrive via token discovery, not documentation.

### Converting passive users into contributors
- Add a **“Good First Fix: Connectors Reliability”** initiative:
  - curated issues (Telegram/Slack/Discord),
  - clear test harness instructions,
  - mentor rotation (weekly office hour in Discord).
- Create a “**Repro bounty**” program (non-monetary: roles, shout-outs, merge priority) for reports like Sw4pIO’s.

---

## 6) Feedback Collection Improvements

### Current channel effectiveness
- **Discord:** high engagement but **signal-to-noise drops** during price volatility; technical questions can be missed.
- **GitHub Issues:** very actionable when filed (excellent repro quality), but likely under-captures “confused but didn’t file” users.

### Improvements for more structured, actionable feedback
1) **Add a short in-app feedback widget** (Milady/app-core + Cloud dashboard)
   - “What were you trying to do?” + logs export toggle + connector/provider selection.
2) **Standardize issue templates per connector**
   - Required fields: connector, runtime mode (bun/node), polling/webhook, expected vs actual, message delivery counts.
3) **Monthly “Top 10 pain points” survey**
   - Separately sample: self-hosters, cloud users, connector-first users, trading-agent builders.

### Underrepresented segments
- **Non-Discord builders** (teams using Slack/Teams, enterprises) — little direct feedback except the Slack plugin migration PR review notes.
- **Mobile/Android users** — despite recent Android work, minimal user feedback captured here.
- **Cloud paying users** — feedback likely happening privately; needs structured capture (billing + auth issues are high-risk).

---

## Prioritized High-Impact Actions (next 2–4 weeks)

1) **Make Telegram “reliably replies 10/10 times” a release gate**  
   - Single poller enforcement + runtime dispatch unification + CI test.

2) **Fix auth/subscription detection regressions with loud failures and migration helpers**  
   - Anthropic stealth preload validation + Codex tokens shape support + consolidated Auth Matrix doc.

3) **Add plugin-sql boot-time schema validation + fresh-db smoke test**  
   - Prevent future “chat unusable on first run” failures.

4) **Stabilize Cloud monetized chat correctness (auth codes + billing invariants)**  
   - Ensure 401/403 classification, reconciliation robustness, and ledger tests.

5) **Reduce Discord support friction during volatility**  
   - Role-based URL filtering + split/support forum channel + bot reporting workflow to protect technical Q&A throughput.