fix: move accordion cells to bottom of files for correct render order in static export

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: ccerv1

notebooks/insights/defi-builder-journeys.py

@@ -1327,52 +1327,6 @@ def transform_pipeline_composition(
     return (pipeline_by_eco,)
 
 
-@app.cell(hide_code=True)
-def section_conclusion(mo, pipeline_by_eco):
-    # pipeline_by_eco dependency ensures this renders after all content cells
-    mo.accordion({
-        "Metrics & Definitions": mo.md("""
-        - Ecosystem Classification:
-            - Ethereum: Projects with >80% of TVL on Ethereum (L1 + L2s like Arbitrum, Polygon, Base, Optimism, etc.)
-            - Solana: Projects whose top chain by TVL is Solana
-            - Other: Remaining non-Ethereum, non-Solana projects (including Hyperliquid, Bitcoin, etc.)
-        - Project Classifications:
-            - Home Project: The DeFi project a builder is primarily associated with, based on their most active repository contributions.
-            - Feeder Projects: Projects where a builder contributed or starred repositories before onboarding to their home DeFi project.
-        - Builder Classifications:
-            - Qualifying Builder: A builder with 12+ months of sustained contribution to a top DeFi project's home repositories.
-            - Monthly Active Builders (MABs): Count of unique qualified builders with at least one contribution event in a given month.
-            - Newcomer: A qualified builder with fewer than 6 months of observable open-source activity before onboarding to their home project (came in cold).
-            - Cohort Retention: The percentage of builders from a given onboarding cohort (year) who remain active on their home project at each subsequent time interval.
-        - Lifecycle:
-            - Onboarding: A builder's first month of activity on their home project.
-            - Offboarding: A builder is considered offboarded if they have been inactive on their home project for 6+ consecutive months.
-            - Still Active: A builder who has contributed to their home project within the most recent 6 months of data.
-        - Flows:
-            - Net Flow: The year-over-year difference between builders entering and exiting an ecosystem. Positive means net talent gain.
-            - Cross-Ecosystem Flow (Ethereum <-> Other Crypto): The horizontal bar in the Net Flows section counts unique builders who moved between Ethereum and other crypto ecosystems over the full time period. This uses a different method from the annual inflow/outflow bars, which count year-over-year flows by source/destination category; the two methods do not sum to each other.
-            - Year-over-Year Flows: The annual bar chart breaks down builder inflows (entering) and outflows (leaving) by partner category (Other Crypto, Non-Crypto OSS, Inactive). The net flow line shows the cumulative balance; these year-over-year counts sum to the total net flow shown in the stat cards.
-        """),
-        "Assumptions & Limitations": mo.md("""
-        - TVL as inclusion criterion: We use TVL to identify economically meaningful protocols, which excludes low-stakes forks and testnet-only experiments; this analysis focuses on capital-securing DeFi.
-        - Excluded protocols: Some high-TVL protocols have minimal observable OSS activity, so they may appear in tables but contribute little to flow analysis.
-        - Private repositories: Activity in private repos is not visible to GitHub Archive, so teams that develop behind closed doors (or move to private repos after open-source phases) may appear to go inactive; this is one of the most consequential limitations.
-        - Post-hire behavior: Some builders are hired by crypto firms and stop public contributions, so the inactive count may be inflated; we cannot distinguish left crypto from went private.
-        - Bot and noise filtering: We exclude known bots, but some automated contributions may still slip through.
-        - Identity resolution: OpenDevData maps multiple accounts to canonical IDs, but imperfect mapping may double-count some builders.
-        - Scope: This is a structured analysis of visible OSS builder flows across economically significant DeFi protocols, not a census of all crypto builders or proprietary development.
-        """),
-        "Data Sources": mo.md("""
-        - [DefiLlama](https://defillama.com/) --- Top DeFi protocols by TVL (40 selected)
-        - [OSS Directory](https://github.com/opensource-observer/oss-directory) --- Protocol to GitHub mapping
-        - [OpenDevData (Electric Capital)](https://github.com/electric-capital/crypto-ecosystems) --- Ecosystem classifications
-        - [GitHub Archive](https://www.gharchive.org/) --- Builder activity events
-        - [OSO API](https://docs.oso.xyz/) --- Data pipeline and metrics
-        """),
-    })
-    return
-
-
 @app.cell(hide_code=True)
 def settings_color_palette():
     # Color palette — semantic colors reused across all charts
@@ -1460,6 +1414,51 @@ def import_libraries():
     return go, make_subplots, pd
 
 
+@app.cell(hide_code=True)
+def section_conclusion(mo):
+    mo.accordion({
+        "Metrics & Definitions": mo.md("""
+        - Ecosystem Classification:
+            - Ethereum: Projects with >80% of TVL on Ethereum (L1 + L2s like Arbitrum, Polygon, Base, Optimism, etc.)
+            - Solana: Projects whose top chain by TVL is Solana
+            - Other: Remaining non-Ethereum, non-Solana projects (including Hyperliquid, Bitcoin, etc.)
+        - Project Classifications:
+            - Home Project: The DeFi project a builder is primarily associated with, based on their most active repository contributions.
+            - Feeder Projects: Projects where a builder contributed or starred repositories before onboarding to their home DeFi project.
+        - Builder Classifications:
+            - Qualifying Builder: A builder with 12+ months of sustained contribution to a top DeFi project's home repositories.
+            - Monthly Active Builders (MABs): Count of unique qualified builders with at least one contribution event in a given month.
+            - Newcomer: A qualified builder with fewer than 6 months of observable open-source activity before onboarding to their home project (came in cold).
+            - Cohort Retention: The percentage of builders from a given onboarding cohort (year) who remain active on their home project at each subsequent time interval.
+        - Lifecycle:
+            - Onboarding: A builder's first month of activity on their home project.
+            - Offboarding: A builder is considered offboarded if they have been inactive on their home project for 6+ consecutive months.
+            - Still Active: A builder who has contributed to their home project within the most recent 6 months of data.
+        - Flows:
+            - Net Flow: The year-over-year difference between builders entering and exiting an ecosystem. Positive means net talent gain.
+            - Cross-Ecosystem Flow (Ethereum <-> Other Crypto): The horizontal bar in the Net Flows section counts unique builders who moved between Ethereum and other crypto ecosystems over the full time period. This uses a different method from the annual inflow/outflow bars, which count year-over-year flows by source/destination category; the two methods do not sum to each other.
+            - Year-over-Year Flows: The annual bar chart breaks down builder inflows (entering) and outflows (leaving) by partner category (Other Crypto, Non-Crypto OSS, Inactive). The net flow line shows the cumulative balance; these year-over-year counts sum to the total net flow shown in the stat cards.
+        """),
+        "Assumptions & Limitations": mo.md("""
+        - TVL as inclusion criterion: We use TVL to identify economically meaningful protocols, which excludes low-stakes forks and testnet-only experiments; this analysis focuses on capital-securing DeFi.
+        - Excluded protocols: Some high-TVL protocols have minimal observable OSS activity, so they may appear in tables but contribute little to flow analysis.
+        - Private repositories: Activity in private repos is not visible to GitHub Archive, so teams that develop behind closed doors (or move to private repos after open-source phases) may appear to go inactive; this is one of the most consequential limitations.
+        - Post-hire behavior: Some builders are hired by crypto firms and stop public contributions, so the inactive count may be inflated; we cannot distinguish left crypto from went private.
+        - Bot and noise filtering: We exclude known bots, but some automated contributions may still slip through.
+        - Identity resolution: OpenDevData maps multiple accounts to canonical IDs, but imperfect mapping may double-count some builders.
+        - Scope: This is a structured analysis of visible OSS builder flows across economically significant DeFi protocols, not a census of all crypto builders or proprietary development.
+        """),
+        "Data Sources": mo.md("""
+        - [DefiLlama](https://defillama.com/) --- Top DeFi protocols by TVL (40 selected)
+        - [OSS Directory](https://github.com/opensource-observer/oss-directory) --- Protocol to GitHub mapping
+        - [OpenDevData (Electric Capital)](https://github.com/electric-capital/crypto-ecosystems) --- Ecosystem classifications
+        - [GitHub Archive](https://www.gharchive.org/) --- Builder activity events
+        - [OSO API](https://docs.oso.xyz/) --- Data pipeline and metrics
+        """),
+    })
+    return
+
+
 @app.cell(hide_code=True)
 def setup_pyoso():
     # This code sets up pyoso to be used as a database provider for this notebook

notebooks/insights/developer-report-2025.py

@@ -10,30 +10,6 @@ def header_title(mo):
     return
 
 
-@app.cell(hide_code=True)
-def header_accordion(mo, _report_complete):
-    mo.accordion({
-        "Metrics & Definitions": mo.md("""
-    - **Time period**: January 2015 to December 2025 (full historical data)
-    - **Monthly Active Developer (MAD)**: A developer who authored at least 1 commit in a given month (measured using a 28-day rolling activity window)
-    - **Tenure Categories**: Newcomers (< 1 year active), Emerging (1–2 years), Established (2+ years)
-    - **Activity Levels**: Full-time (sustained activity over 84-day window), Part-time (intermittent), One-time (sporadic)
-        """),
-        "Assumptions & Limitations": mo.md("""
-    - **Commit-only activity measure**: Developer activity is based on commits only — pull requests, code reviews, and issue comments are not counted
-    - **Public repos only**: Private repositories are excluded from the dataset
-    - **Identity resolution**: Developer identities are resolved across forges using Electric Capital's methodology, but some cross-account connections may be missed
-    - **Ecosystem classification**: Ecosystem assignments follow the Electric Capital taxonomy; projects may belong to multiple ecosystems
-        """),
-        "Data Sources": mo.md("""
-    - **Open Dev Data** — Electric Capital's developer activity dataset, [github.com/electric-capital/crypto-ecosystems](https://github.com/electric-capital/crypto-ecosystems)
-    - **OSO API** — Data pipeline and metrics, [docs.oso.xyz](https://docs.oso.xyz/)
-    - **Key Models** — `oso.stg_opendevdata__eco_mads`, `oso.stg_opendevdata__ecosystems`
-        """),
-    })
-    return
-
-
 @app.cell(hide_code=True)
 def setup_imports():
     import pandas as pd
@@ -1634,8 +1610,31 @@ def comparison_chart(df_all, mo, pd):
     )
     _src = _html_mod.escape(_inner, quote=True)
     mo.Html(f'<iframe srcdoc="{_src}" class="ddp-chart-frame" scrolling="no"></iframe>')
-    _report_complete = True
-    return (_report_complete,)
+    return
+
+
+@app.cell(hide_code=True)
+def header_accordion(mo):
+    mo.accordion({
+        "Metrics & Definitions": mo.md("""
+    - **Time period**: January 2015 to December 2025 (full historical data)
+    - **Monthly Active Developer (MAD)**: A developer who authored at least 1 commit in a given month (measured using a 28-day rolling activity window)
+    - **Tenure Categories**: Newcomers (< 1 year active), Emerging (1–2 years), Established (2+ years)
+    - **Activity Levels**: Full-time (sustained activity over 84-day window), Part-time (intermittent), One-time (sporadic)
+        """),
+        "Assumptions & Limitations": mo.md("""
+    - **Commit-only activity measure**: Developer activity is based on commits only — pull requests, code reviews, and issue comments are not counted
+    - **Public repos only**: Private repositories are excluded from the dataset
+    - **Identity resolution**: Developer identities are resolved across forges using Electric Capital's methodology, but some cross-account connections may be missed
+    - **Ecosystem classification**: Ecosystem assignments follow the Electric Capital taxonomy; projects may belong to multiple ecosystems
+        """),
+        "Data Sources": mo.md("""
+    - **Open Dev Data** — Electric Capital's developer activity dataset, [github.com/electric-capital/crypto-ecosystems](https://github.com/electric-capital/crypto-ecosystems)
+    - **OSO API** — Data pipeline and metrics, [docs.oso.xyz](https://docs.oso.xyz/)
+    - **Key Models** — `oso.stg_opendevdata__eco_mads`, `oso.stg_opendevdata__ecosystems`
+        """),
+    })
+    return
 
 
 if __name__ == "__main__":

notebooks/insights/developer-retention.py

@@ -12,37 +12,6 @@ def header_title(mo):
 
 
 
-@app.cell(hide_code=True)
-def header_accordion(mo):
-    mo.accordion({
-        "Metrics & Definitions": mo.md("""
-**Definitions**
-
-- **Cohort**: Developers grouped by the year of their first contribution to the ecosystem
-- **Retention Rate**: Percentage of the original cohort that remains active in subsequent periods
-- **Years Since Join**: Time elapsed since first contribution (Year 0 = joined year, always 100%)
-
-**Methodology**
-
-1. **Cohort Assignment**: Each developer is assigned to a cohort based on their first contribution date
-2. **Activity Tracking**: We track whether the developer had any activity in subsequent years
-3. **Retention Rate**: Percentage of the original cohort that remains active
-        """),
-        "Assumptions & Limitations": mo.md("""
-- **Multi-ecosystem developers**: Developers active in multiple ecosystems are counted separately per ecosystem — a developer who churns from one ecosystem may still be active in another
-- **Identity resolution**: Developer identities are resolved by Electric Capital's fingerprinting; the same person using different accounts may be counted multiple times
-- **Newer cohorts**: More recent cohorts have shorter observation windows and therefore fewer data points for retention analysis
-- **Public commits only**: Only public GitHub activity is tracked; private repos and non-GitHub platforms are excluded
-- **Activity windows**: Activity is measured using 28-day rolling windows via Open Dev Data's `repo_developer_28d_activities` model
-        """),
-        "Data Sources": mo.md("""
-- **Open Dev Data (Electric Capital)** — Developer activity data, [github.com/electric-capital/crypto-ecosystems](https://github.com/electric-capital/crypto-ecosystems)
-- **Key Models** — `oso.stg_opendevdata__repo_developer_28d_activities`, `oso.stg_opendevdata__ecosystems_repos_recursive`
-        """),
-    })
-    return
-
-
 @app.cell(hide_code=True)
 def test_connection(mo, pyoso_db_conn):
     _test_df = mo.sql("""SELECT 1 AS test""", engine=pyoso_db_conn, output=False)
@@ -478,6 +447,37 @@ def apply_ec_style(fig, title=None, subtitle=None, y_title=None, show_legend=Tru
     return (apply_ec_style,)
 
 
+@app.cell(hide_code=True)
+def header_accordion(mo):
+    mo.accordion({
+        "Metrics & Definitions": mo.md("""
+**Definitions**
+
+- **Cohort**: Developers grouped by the year of their first contribution to the ecosystem
+- **Retention Rate**: Percentage of the original cohort that remains active in subsequent periods
+- **Years Since Join**: Time elapsed since first contribution (Year 0 = joined year, always 100%)
+
+**Methodology**
+
+1. **Cohort Assignment**: Each developer is assigned to a cohort based on their first contribution date
+2. **Activity Tracking**: We track whether the developer had any activity in subsequent years
+3. **Retention Rate**: Percentage of the original cohort that remains active
+        """),
+        "Assumptions & Limitations": mo.md("""
+- **Multi-ecosystem developers**: Developers active in multiple ecosystems are counted separately per ecosystem — a developer who churns from one ecosystem may still be active in another
+- **Identity resolution**: Developer identities are resolved by Electric Capital's fingerprinting; the same person using different accounts may be counted multiple times
+- **Newer cohorts**: More recent cohorts have shorter observation windows and therefore fewer data points for retention analysis
+- **Public commits only**: Only public GitHub activity is tracked; private repos and non-GitHub platforms are excluded
+- **Activity windows**: Activity is measured using 28-day rolling windows via Open Dev Data's `repo_developer_28d_activities` model
+        """),
+        "Data Sources": mo.md("""
+- **Open Dev Data (Electric Capital)** — Developer activity data, [github.com/electric-capital/crypto-ecosystems](https://github.com/electric-capital/crypto-ecosystems)
+- **Key Models** — `oso.stg_opendevdata__repo_developer_28d_activities`, `oso.stg_opendevdata__ecosystems_repos_recursive`
+        """),
+    })
+    return
+
+
 @app.cell(hide_code=True)
 def setup_pyoso():
     # This code sets up pyoso to be used as a database provider for this notebook